Diagnosis and treatment of metabolic dysfunction-associated steatotic liver disease (MASLD)

Abstract

Featured are methods of identifying (e.g., diagnosing) metabolic dysfunction-associated steatotic liver disease (MASLD), such as MASH (e.g., at-risk MASH) in a subject by, e.g., determining a level of one or more biomarkers in a biological sample (e.g., blood, serum, or plasma) obtained from the subject. Also featured are methods of treating, monitoring, or evaluating the efficacy of a treatment for MASLD (e.g., MASH, e.g., at-risk MASH) in a subject based on the level of one or more biomarkers in a biological sample (e.g., a blood, serum, or plasma sample) obtained from the subject.

Description

[0001] DIAGNOSIS AND TREATMENT OF METABOLIC DYSFUNCTION-ASSOCIATED STEATOTIC LIVER DISEASE (MASLD)

[0002] BACKGROUND

[0003] Metabolic dysfunction-associated steatotic liver disease (MASLD), also known as non-alcoholic fatty liver disease (NAFLD), affects >30% of the general population and constitutes the primary cause of liver-related morbidity and mortality. Between 10-25% of patients with MASLD progress to metabolic dysfunction-associated steatohepatitis (MASH), also known as non-alcoholic associated steatohepatitis (NASH), further complicated by hepatocellular scarring (e.g., fibrosis), potentially advancing liver decompensation, hepatocellular carcinoma (e.g., cirrhosis), and the risk of cardiometabolic disease. The diagnostic gold standard for MASLD is liver biopsy, an invasive procedure with inherent risk, cost, and sampling considerations. The two preeminent histopathological scoring systems for MASLD are the NAFLD activity score (NAS) and the liver fibrosis score, both of which are fully encompassed under the scoring scheme of the NASH Clinical Research Network (NASH CRN). Notably, MASH accompanied by significant fibrosis (e.g., a fibrosis score of F>2), coined “at-risk MASH”, represents a substantial increase in the likelihood of morbidity and mortality and designates a special population of interest demarcated by the Food and Drug Administration (FDA) as necessitating and benefiting from early treatment.

[0004] Apart from imaging modalities, identification of patients with at-risk MASH stipulates the use of non-invasive blood-based biomarkers principally designed for disease detection, however almost all display suboptimal accuracies and cannot determine the presence and degree of MASH. For the diagnosis of steatosis, ultrasound is broadly used and demonstrates acceptable specificity but suboptimal sensitivity, especially in earlier stages of the disease, while being largely dependent on operator experience. Moreover, optimal thresholds for the controlled attenuation parameter (CAP) measurements (e.g., M or XL-probes) are not universally set and likewise produce suboptimal accuracies. On the other hand, magnetic resonance imaging derived proton density fat fraction (MRI-PDFF), transient elastography, or magnetic resonance elastography (MRE) display higher accuracy for steatosis and fibrosis detection, respectively, but are costly and may not universally available.

[0005] Therefore, there remains a need in the field for non-invasive methods of diagnosing a subject with MASLD (e.g., MASH, e.g., at-risk MASH). For example, diagnosing patients as having F2-F3 liver fibrosis would be especially useful for identifying subjects that are eligible for treatment with resmetirom (REZDIFFRA®), a recently approved MASLD therapy for patients with F2-F3 liver fibrosis.

[0006] SUMMARY OF THE INVENTION

[0007] In one aspect, the disclosure features a method for diagnosing metabolic dysfunction-associated steatotic liver disease (MASLD) in a subject, the method including determining a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject, wherein an increased level (e.g., an increase of 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) of the at least one biomarker relative to a reference is indicative of the presence of MASLD, and wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, S-adenosylhomocysteine (SAH), 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7- FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4- FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine.

[0008] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0009] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0010] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0011] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0012] In some embodiments, the at least one biomarker is kynurenine.

[0013] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0014] In some embodiments, the subject has stage 1 liver fibrosis (F1 ), stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), or stage 4 liver fibrosis (F4).

[0015] In some embodiments, the MASLD is metabolic dysfunction-associated steatotic liver (MAFL) or metabolic dysfunction-associated steatohepatitis (MASH).

[0016] In some embodiments, the MASH is at-risk MASH.

[0017] In some embodiments, the at-risk MASH is MASH F2, MASH F3, MASH F4, MASH F2-F3 (e.g., MASH F2 or MASH F3) or MASH F3-F4 (e.g., MASH F3 or MASH F4), wherein optionally the subject does not have liver cirrhosis.

[0018] In some embodiments, the at-risk MASH is MASH F3-F4 (e.g., MASH F3 or MASH F4), wherein optionally the subject has liver cirrhosis.

[0019] In some embodiments, the subject has a non-alcoholic steatohepatitis activity (NAS) score >4 (e.g., a NAS score of 4, NAS score of 5, NAS score of 6, NAS score of 7, or a NAS score of 8).

[0020] In some embodiments, the level of the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) relative to the reference and the method further includes administering a MASLD therapy to the subject.

[0021] In some embodiments, the disclosure features a method of treating MASLD in a subject determined to have MASLD as described in the foregoing embodiments, where the method includes administering a MASLD therapy to the subject.

[0022] In some embodiments, the disclosure features a method of treating MASH in a subject determined to have MASH as described in the foregoing embodiments, where the method includes administering a MASH therapy to the subject. In another aspect, the disclosure features a method of treating MASLD in a subject, the method including: (a) identifying the subject as having MASLD by determining that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject is increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering a MASLD therapy to the subject.

[0023] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0024] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0025] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0026] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0027] In some embodiments, the at least one biomarker is kynurenine.

[0028] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0029] In some embodiments, the subject has stage 1 liver fibrosis (F1 ), stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), or stage 4 liver fibrosis (F4).

[0030] In some embodiments, the MASLD is MAFL or MASH.

[0031] In some embodiments, the MASH is at-risk MASH.

[0032] In some embodiments, the at-risk MASH is MASH F2, MASH F3, MASH F4, MASH F2-F3 (e.g., MASH F2 or MASH F3) or MASH F3-F4 (e.g., MASH F3 or MASH F4), wherein optionally the subject does not have liver cirrhosis.

[0033] In some embodiments, the at-risk MASH is MASH F3-F4 (e.g., MASH F3 or MASH F4), wherein optionally the subject has liver cirrhosis.

[0034] In some embodiments, the subject has a NAS score >4 (e.g., a NAS score of 4, NAS score of 5, NAS score of 6, NAS score of 7, or a NAS score of 8).

[0035] In some aspects, the at least one biomarker is increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference.

[0036] In some embodiments, the method further includes a step of monitoring treatment efficacy of the MASLD therapy administered to the subject.

[0037] In some embodiments, treatment efficacy is monitored once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily.

[0038] In some embodiments, the step of monitoring treatment efficacy includes: (a) identifying the subject’s condition as: (I) progressive, wherein the level of the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more additional biomarkers) in a biological sample obtained from the subject has increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) over time relative to a reference level or a previous sample obtained from the subject; or (ii) stabilized, wherein the level of the at least one biomarker in a biological sample obtained from the subject has remained unchanged (e.g., an increase or decrease <5%, relative to a reference) over time relative to a reference level or a previous sample obtained from the subject; and (b) modifying the MASLD therapy administered to the subject.

[0039] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0040] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4. In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a-ketoglutarate.

[0041] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0042] In some embodiments, the at least one biomarker is kynurenine.

[0043] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0044] In some embodiments, modifying the MASLD therapy administered to the subject includes increasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly) of a therapeutic agent to the subject.

[0045] In some embodiments, the dosage is increased by at least 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly.

[0046] In some embodiments, the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

[0047] In some embodiments, the step of monitoring treatment efficacy includes: (a) identifying the subject’s condition as in remission by determining that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject has decreased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) over time relative to a reference level or a previous sample obtained from the subject; and (b) modifying the MASLD therapy administered to the subject.

[0048] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0049] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0050] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0051] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0052] In some embodiments, the at least one biomarker is kynurenine.

[0053] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0054] In some embodiments, modifying the MASLD therapy administered to the subject includes decreasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., decreased to 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years) of a therapeutic agent to the subject.

[0055] In some embodiments, the dosage is decreased by at least 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years.

[0056] In some embodiments, the level of the at least one biomarker has decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

[0057] In another aspect, the disclosure includes a method of modifying a MASLD therapy in a subject including: increasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., increased to at least, 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly) of a MASLD therapy in the subject, where the subject is determined to have an increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) or unchanged level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, where the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'- deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O- methyluridine; and further administering the modified MASLD therapy to the subject.

[0058] In some embodiments, the at least one biomarker is selected from the group consisting of: 3- ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0059] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0060] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0061] In some embodiments, the at least one biomarker is kynurenine.

[0062] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0063] In some embodiments, the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

[0064] In some embodiments, modifying the MASLD therapy administered to the subject includes increasing the dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., increased to at least, 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly) of the MASLD therapy to the subject.

[0065] In one aspect, the disclosure includes a method of modifying a MASLD therapy in a subject, which includes first decreasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., decreased to 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years) of a MASLD therapy in the subject, wherein the subject is determined to have an decreased level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine; and further administering the modified MASLD therapy to the subject.

[0066] In some embodiments, modifying the MASLD therapy administered to the subject includes decreasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., decreased to 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years) of a therapeutic agent to the subject.

[0067] In another aspect, the disclosure features a method of monitoring treatment efficacy in a subject being treated with a MASLD therapy, the method including: (a) detecting an increase in a level (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3- indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 - stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0- FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4-hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s- sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the MASLD therapy administered to the subject.

[0068] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0069] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0070] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate

[0071] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0072] In some embodiments, the at least one biomarker is kynurenine.

[0073] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0074] In some embodiments, the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

[0075] In some embodiments, modifying the MASLD therapy administered to the subject includes increasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration of a therapeutic agent to the subject.

[0076] In another aspect, the disclosure features a method of monitoring treatment efficacy in a subject (e.g., by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily) being treated with a MASLD therapy, the method including: (a) detecting a decrease (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a- ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N- acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the MASLD therapy administered to the subject.

[0077] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0078] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0079] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0080] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0081] In some embodiments, the at least one biomarker is kynurenine.

[0082] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0083] In some embodiments, modifying the MASLD therapy administered to the subject includes decreasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., increased to at least, 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly) of a therapeutic agent to the subject.

[0084] In some embodiments, the subject has stage 1 liver fibrosis (F1 ), stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), or stage 4 liver fibrosis (F4).

[0085] In some embodiments, the subject has MASLD.

[0086] In some embodiments, the MASLD is MAFL or MASH.

[0087] In some embodiments, the MASH is at-risk MASH.

[0088] In some embodiments, the at-risk MASH is MASH F2-F3 (e.g., MASH F2 or MASH F3), wherein optionally the subject does not have liver cirrhosis.

[0089] In some embodiments, the at-risk MASH is MASH F3-F4 (e.g., MASH F3 or MASH F4), wherein optionally the subject has liver cirrhosis. In some embodiments, the subject has a NAS score >4 (e.g., a NAS score of 4, NAS score of 5, NAS score of 6, NAS score of 7, or a NAS score of 8).

[0090] In another aspect, the disclosure features a method of monitoring (e.g., once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily) disease progression in a subject with MASH, the method including: (a) detecting an increase in a level (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the increase in the level of the at least one biomarker indicates that MASH has progressed to at-risk MASH, and wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'- deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O- methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering a MASLD therapy to said subject.

[0091] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0092] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0093] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0094] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0095] In some embodiments, the at least one biomarker is kynurenine.

[0096] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0097] In some embodiments, the MASLD therapy is weight loss, a dietary change, increased physical activity, a bariatric surgery, an anti-inflammatory agent, an anti-apoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a peroxisome proliferator-activated receptor (PPAR) modulator, a thyroid receptor beta agonist, an incretin receptor agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a farnesoid X receptor (FXR) agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), JKB-121 , and / or resmetirom (REZDIFFRA™).

[0098] In some embodiments, the MASLD therapy is resmetirom (REZDIFFRA™).

[0099] In some embodiments, the thyroid beta receptor agonist is resmetirom (REZDIFFRA™). In some embodiments, the method reduces or delays the progression of the subject’s MASLD (e.g., by at least 1 day, 5 days, 10 days, 20 days, 1 month, 2 months, 3 months, 4 months, 5 months, 6 months, 7 months 8 months, 9 months, 10 months, 11 months, 1 year, 1.5 years, 2 years, 2.5 years, 3 years, 3.5 years, 4 years, 5 years, 6 years, 7 years, 8 years, 9 years, 10 years, 15 years, 20 years, 25 years, 30 years, 35 years, 40 years, 50 years, 55 years, or more).

[0100] In another aspect, the disclosure features a method for diagnosing liver fibrosis in a subject, the method including determining a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject, wherein an increased level (e.g., an increase of 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) of the at least one biomarker relative to a reference is indicative of the presence of liver fibrosis, and wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X- 26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2- methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4- FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5- cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4-hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine.

[0101] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0102] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0103] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0104] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0105] In some embodiments, the at least one biomarker is kynurenine.

[0106] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0107] In some embodiments, the liver fibrosis is stage 2 liver fibrosis (F2) or stage 3 liver fibrosis (F3).

[0108] In some embodiments, the liver fibrosis is stage 3 liver fibrosis (F3) or stage 4 liver fibrosis (F4). In some embodiments, the liver fibrosis is F2-F3. In some embodiments, the liver fibrosis is F3-F4.

[0109] In some embodiments, the level of the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) relative to the reference and the method further includes administering a liver fibrosis therapy to the subject. In another aspect, the disclosure includes a method of treating liver fibrosis in a subject determined to have liver fibrosis as described in the foregoing embodiments, where the method includes administering a liver fibrosis therapy to the subject.

[0110] In another aspect, the disclosure features a method of treating liver fibrosis in a subject, the method including: (a) identifying the subject as having liver fibrosis by determining that a level of at least one biomarker (e.g. , one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject is increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a- ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N- acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering a liver fibrosis therapy to the subject.

[0111] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0112] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0113] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0114] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0115] In some embodiments, the at least one biomarker is kynurenine.

[0116] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0117] In some embodiments, the liver fibrosis is F2, F3, or F2-F3.

[0118] In some embodiments, the liver fibrosis is F3, F4, or F3-F4.

[0119] In some embodiments, the subject has a NAS score >4 (e.g., a NAS score of 4, NAS score of 5, NAS score of 6, NAS score of 7, or a NAS score of 8).

[0120] In some embodiments, the method further includes a step of monitoring treatment efficacy of the liver fibrosis therapy administered to the subject (e.g., by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily).

[0121] In some embodiments, the step of monitoring treatment efficacy (e.g., by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily) includes: (a) identifying the subject’s condition as: (i) progressive by determining that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject has increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) over time relative to a reference level or a previous sample obtained from the subject; or (ii) stabilized by determining that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject has remained unchanged (e.g., an increase or decrease <5%, relative to a reference) over time relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3- ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTLJIL, 2'- deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O- methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the liver fibrosis therapy administered to the subject.

[0122] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine.

[0123] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTLJIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0124] In some embodiments, the at least one biomarker (e.g., one, two, three, four, or five biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0125] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0126] In some embodiments, the at least one biomarker is 3-ureidopropionate. In some embodiments, the at least one biomarker is kynurenine.

[0127] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0128] In some embodiments, modifying the liver fibrosis therapy administered to the subject includes increasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly) of a therapeutic agent to the subject.

[0129] In some embodiments, the step of monitoring (e.g., by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily) treatment efficacy includes: (a) identifying the subject’s condition as in remission by determining that a level of at least one additional biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject has decreased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) over time relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the liver fibrosis therapy administered to the subject.

[0130] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0131] In some embodiments, modifying the liver fibrosis therapy administered to the subject includes decreasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years) of a therapeutic agent to the subject.

[0132] In some embodiments, the at least one additional biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4. In some embodiments, the at least one additional biomarker is 3-ureidopropionate, kynurenine, and / or a-ketoglutarate.

[0133] In some embodiments, the at least one additional biomarker is 3-ureidopropionate.

[0134] In some embodiments, the at least one additional biomarker is kynurenine.

[0135] In some embodiments, the at least one additional biomarker is a-ketoglutarate.

[0136] In another aspect, the disclosure features a method of monitoring (e.g., by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily) treatment efficacy in a subject being treated with a liver fibrosis therapy, the method including: (a) detecting an increase (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the liver fibrosis therapy administered to the subject.

[0137] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0138] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0139] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0140] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0141] In some embodiments, the at least one biomarker is kynurenine.

[0142] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0143] In some embodiments, modifying the liver fibrosis therapy administered to the subject includes increasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly) of a therapeutic agent to the subject. In another aspect, the disclosure features a method of monitoring treatment efficacy in a subject being treated with a liver fibrosis therapy, the method including: (a) detecting a decrease (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the liver fibrosis therapy administered to the subject.

[0144] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0145] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0146] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0147] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0148] In some embodiments, the at least one biomarker is kynurenine.

[0149] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0150] In some embodiments, modifying the liver fibrosis therapy administered to the subject includes decreasing a dosage (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more) and / or frequency of administration (e.g., decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years) of a therapeutic agent to the subject.

[0151] In some embodiments, the subject has F2 or F3 liver fibrosis. In some embodiments, the subject does not have liver cirrhosis.

[0152] In some embodiments, the subject has F3 or F4 liver fibrosis. In some embodiments, the subject does not have liver cirrhosis. In some embodiments, the subject has liver cirrhosis. In some embodiments, the method reduces or delays the progression of the subject’s fibrosis.

[0153] In another aspect, the disclosure features a method of monitoring (e.g., by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily) disease progression in a subject with liver fibrosis, the method including: (a) detecting an increase (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the increase in the level of the at least one biomarker indicates that the liver fibrosis has progressed to stage F2, F3, or F4 liver fibrosis, and wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering a liver fibrosis therapy to said subject.

[0154] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0155] In some embodiments, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0156] In some embodiments, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0157] In some embodiments, the at least one biomarker is 3-ureidopropionate.

[0158] In some embodiments, the at least one biomarker is kynurenine.

[0159] In some embodiments, the at least one biomarker is a-ketoglutarate.

[0160] In some embodiments, the liver fibrosis therapy is or includes weight loss, a dietary change, increased physical activity, a bariatric surgery, a liver transplant an ani-fibrotic agent, an anti-inflammatory agent, an anti-apoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a PPAR agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a FXR agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, CVC, JKB-121 , or resmetirom (REZDIFFRA™).

[0161] In some embodiments, the liver fibrosis therapy is resmetirom (REZDIFFRA™).

[0162] In some embodiments, the thyroid beta receptor agonist is resmetirom (REZDIFFRA™). In some embodiments, the liver fibrosis has progressed from F2 to F3.

[0163] In some embodiments, the liver fibrosis has progressed from F3 to F4.

[0164] In yet another aspect, the disclosure features a method of stratifying a subject having MASLD for treatment including: (a) determining a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level; and (b) classifying the status of the subject’s MASLD as at-risk MASH based on the level determined in step (a). The level of the at least one biomarker determined according to the method may be a change (e.g., an increase or a decrease) relative to the reference level.

[0165] In some embodiments, an increase in the level of the at least one biomarker relative to the reference level classifies the status of the subject’s MASLD as at-risk MASH, wherein the at least one biomarker selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'- O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleg lyoxylic acid, 3-ureidopropionate, 4- hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 - phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2- FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and / or (b) a decrease in the level of the at least one biomarker relative to the reference level classifies the status of the subject’s MASLD as at-risk MASH, wherein the at least one biomarker selected from the group including of: 2'-deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.

[0166] In some embodiments, the method of stratifying a subject having MASLD for treatment further includes classifying the subject as a candidate or non-candidate for receiving a MASLD therapy based on the status of the subject’s MASLD, wherein: (a) the subject’s MASLD is classified as at-risk MASH, thereby classifying the subject as a candidate to receive at least one MASLD therapy selected from the group including of: an anti-inflammatory agent, an anti-apoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a peroxisome proliferator- activated receptor (PPAR) modulator, a thyroid receptor beta agonist, an incretin receptor agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a farnesoid X receptor (FXR) agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), JKB-121 , and resmetirom (REZDIFFRA™); or (b) the subject’s MASLD is not classified as at-risk MASH, thereby classifying the subject as a non-candidate for resmetirom (REZDIFFRA™).

[0167] In some embodiments of any of the above aspects, the method further includes evaluating one or more additional parameters selected from the group consisting of liver transaminases, platelets, age, BMI, and PNPLA3 genotype. In some embodiments of any of the above aspects, the method further includes using one or more different non-invasive tests (NITs) known in the art (e.g., the NITs tested against the predictive model of the present disclosure as shown in FIG. 14B), wherein, optionally, the one or more different NITs confirm the disease state (e.g., MASLD, MASH, at risk MASH) and / or stage (e.g., F2, F3, F4). In some embodiments of any of the above aspects, the reference level is established by performing classifier training using 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and / or inosine levels obtained from subjects clinically diagnosed as having MASH, subjects clinically diagnosed as having fibrosis, and healthy subjects.

[0168] In some embodiments of any of the above aspects, the reference level is established by performing classifier training using 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, or 2'-O-methyluridine levels obtained from subjects clinically diagnosed as having MASH, subjects clinically diagnosed as having fibrosis, and healthy subjects.

[0169] In some embodiments of any of the above aspects, the reference level is established by performing classifier training using 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, or TAG54:4 / FA22:4 levels obtained from subjects clinically diagnosed as having MASH, subjects clinically diagnosed as having fibrosis, and healthy subjects.

[0170] In some embodiments of any of the above aspects, the reference level is established by performing classifier training using 3-ureidopropionate, kynurenine, and / or a-ketoglutarate levels obtained from subjects clinically diagnosed as having MASH, subjects clinically diagnosed as having fibrosis, and healthy subjects.

[0171] In some embodiments of any of the above aspects, the reference level is established by training a machine learning algorithm using a reference dataset.

[0172] In some embodiments of any of the above aspects, the level of the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) is determined by one or more of mass spectrometry (MS), liquid chromatography-MS (LC / MS), gas chromatography (GO), or enzyme linked immunosorbent assay (ELISA).

[0173] In some embodiments of any of the above aspects, the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) includes at least two biomarkers, at least three biomarkers, at least four biomarkers, or at least five biomarkers (e.g., five, six, seven, eight, nine, ten, or more biomarkers). In some embodiments: (a) the at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) includes:(i) 3-ureidopropionate; (ii) kynurenine; (ill) a- ketoglutarate; (iv) mannonate; or (v) TAG54:4 / FA22:4, (b) the at least two biomarkers include: (i) 3- ureidopropionate and kynurenine; (II) 3-ureidopropionate and a-ketoglutarate; (ill) 3-ureidopropionate and mannonate; (iv) 3-ureidopropionate and TAG54:4 / FA22:4; (v) kynurenine and a-ketoglutarate; (vi) kynurenine and mannonate; (vii) kynurenine and TAG54:4 / FA22:4; (viii) a-ketoglutarate and mannonate; (ix) a- ketoglutarate and TAG54:4 / FA22:4; or (x) mannonate and TAG54:4 / FA22:4, (c) the at least three biomarkers include: (i) 3-ureidopropionate, kynurenine, and a- ketoglutarate; (ii) 3-ureidopropionate, kynurenine, and mannonate; (iii) 3-ureidopropionate, kynurenine, and TAG54:4 / FA22:4; (iv) 3- ureidopropionate, a-ketoglutarate, and mannonate; (v) 3-ureidopropionate, a-ketoglutarate, and TAG54:4 / FA22:4; (vi) 3-ureidopropionate, mannonate, and TAG54:4 / FA22:4; (vii) kynurenine, a- ketoglutarate, and mannonate; (viii) kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4; (ix) kynurenine, mannonate, and TAG54:4 / FA22:4; or (x) a-ketoglutarate, mannonate, and TAG54:4 / FA22:4, (d) the at least four biomarkers include: (i) kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4; (ii) 3- ureidopropionate, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4; (iii) 3-ureidopropionate, kynurenine, mannonate, and TAG54:4 / FA22:4; (iv) 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4; or (v) 3-ureidopropionate, kynurenine, a-ketoglutarate, and mannonate, or (e) the at least five biomarkers include 3-ureidopropionate kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0174] In some embodiments: (a) the at least one biomarker includes: (i) 3-ureidopropionate; (ii) kynurenine; or (iii) a-ketoglutarate; (b) the at least two biomarkers include: (i) 3-ureidopropionate and kynurenine; (ii) 3-ureidopropionate and a-ketoglutarate; or (iii) kynurenine and a-ketoglutarate; or (c) the at least three biomarkers include 3-ureidopropionate, kynurenine, and a-ketoglutarate.

[0175] In some embodiments, the at least one biomarker, at least two biomarkers, or at least three biomarkers further comprises an additional biomarker selected from the group consisting of: X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3- indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0176] In some embodiments, the at least one biomarker, at least two biomarkers, at least three biomarkers, at least four biomarkers, or at least five biomarkers further includes an additional biomarker selected from the group consisting of: X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine.

[0177] In some embodiments, the at least one biomarker, at least two biomarkers, at least three biomarkers, at least four biomarkers, or at least five biomarkers further includes an additional biomarker selected from the group consisting of: X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2- methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

[0178] In some embodiments, the biological sample is a whole blood, serum, or plasma sample.

[0179] In some embodiments, the biological sample is a serum sample.

[0180] In some embodiments, the subject has a body mass index (BMI) less than 27.5 kg / m2. In some embodiments, the subject has a BMI greater than or equal to 27.5 kg / m2. In another aspect, the disclosure features a MASLD therapy for use in treatment of a subject having MASLD, the treatment including: (a) identifying the subject as having MASLD by determining that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject is increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 3- ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'- deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2' 0- methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering the MASLD therapy to the subject.

[0181] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0182] In another aspect, the disclosure features a MASLD therapy for use in monitoring efficacy of treatment in a subject having MASLD, the treatment including: (a) detecting an increase (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the MASLD therapy administered to the subject.

[0183] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0184] In another aspect, the disclosure features a MASLD therapy for use in monitoring efficacy of a treatment in a subject having MASLD, the treatment including: (a) detecting a decrease (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the MASLD therapy administered to the subject.

[0185] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0186] In another aspect, the disclosure features a MASLD therapy for use in a method of monitoring disease progression in a subject having MAFL, the method including: (a) detecting an increase (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the increase in the level of the at least one biomarker indicates that MAFL has progressed to MASH, and wherein the at least one biomarker is selected from the group consisting of 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X- 26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2- methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4- FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5- cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4-hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering the MASLD therapy to said subject.

[0187] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0188] In another aspect, the disclosure features a liver fibrosis therapy for use in treatment of a subject having liver fibrosis, the treatment including: (a) identifying the subject as having liver fibrosis by determining that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject is increased (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 3- ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'- deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O- methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering the liver fibrosis therapy to the subject.

[0189] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0190] In another aspect, the disclosure features a liver fibrosis therapy for use in monitoring efficacy of treatment in a subject having liver fibrosis, the treatment including: (a) detecting an increase (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the liver fibrosis therapy administered to the subject.

[0191] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0192] In another aspect, the disclosure features a liver fibrosis therapy for use in monitoring efficacy of treatment in a subject having liver fibrosis, the treatment including: (a) detecting a decrease (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyltransferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10- FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) modifying the liver fibrosis therapy administered to the subject.

[0193] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0194] In another aspect, the disclosure features a liver fibrosis therapy for use in a method of monitoring disease progression in a subject having liver fibrosis, the method including: (a) detecting an increase (e.g., by 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more, relative to a reference) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the increase in the level of the at least one biomarker indicates that the liver fibrosis has progressed to stage F2, F3, or F4 liver fibrosis, and wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleg lyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, 2'-O-methyluridine, gamma glutamyl-transferase, pantothenate, succinate, TAG45:1 -FA15:0, dimethylarginine (SDMA + ADMA), quinolinate, DCER(18:0), TAG56:4-FA16:0, TAG44:1 -FA18:1 , 1 -stearoyl-GPG (18:0), TAG58:7-FA22:5, TAG60:10-FA22:6, sphingadienine, leucylalanine, TAG52:0-FA20:0, DAG(18:1 / 22:5), TAG54:4-FA18:1 , 3beta,7alpha-dihydroxy-5-cholestenoate, TAG50:2-FA16:0, threonylphenylalanine, 4- hydroxychlorothalonil, DAG(18:1 / 20:2), sphingosine 1 -phosphate, cysteine s-sulfate, TAG50:4-FA18:2, TAG48:0-FA14:0, TAG47:0-FA15:0, and inosine; and (b) administering the liver fibrosis therapy to said subject.

[0195] In an embodiment, the at least one biomarker is 3-ureidopropionate, kynurenine, and / or a- ketoglutarate.

[0196] In another aspect, the disclosure features a computer implemented method for detection of a MASLD disease state in a subject including: (a) identifying one or more biomarkers associated with MASLD; (b) training a predictive computer model based on values of one or more biomarkers; (c) receiving information regarding a subject through a graphical user interface, wherein the information includes the level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from a subject; (d) determining the risk probability score of the disease state in the subject by applying the predictive computer model; and (e) automatically displaying the risk probability score on the graphical user interface.

[0197] In another aspect, the disclosure features a computer implemented method for detection of a MASLD disease state in a subject including: (a) inputting a level of at least one biomarker selected from the group consisting of: 3-ureidopropionate, a-ketoglutarate, and kynurenine, determined in a biological sample obtained from a subject; (b) determining a risk probability score of the disease state in the subject by applying a predictive computer model previously trained using verified liver disease reference cases; and (c) automatically displaying the risk probability score on a graphical user interface. In another aspect, the disclosure features a computer implemented method for detection of

[0198] MASLD in a subject including: (a) identifying one or more biomarkers associated with MASLD; (b) training a predictive computer model based on values of one or more biomarkers; (c) receiving information regarding a subject through a graphical user interface, wherein the information includes the level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from a subject; (d) determining the risk probability score of the disease state in the subject by applying the predictive computer model; and (e) automatically displaying the risk probability score on the graphical user interface.

[0199] In another aspect, the disclosure features a computer implemented method for detection of MASLD in a subject including: (a) inputting a the level of at least one biomarker selected from the group consisting of: 3-ureidopropionate, a-ketoglutarate, and kynurenine, determined in a biological sample obtained from a subject; (b) determining a risk probability score of MASLD in the subject by applying a predictive computer model previously trained using verified liver disease reference cases; and (c) automatically displaying the risk probability score on a graphical user interface.

[0200] In another aspect, the disclosure features a diagnostic device for determing the risk of MASLD disease progression in a subject including: (a) means to acquire data including a level of one or more biomarkers; (b) optionally, an analysis module operable to derive corrections of the data; (c) output means for producing a statistical model with the data including at least one metric of the level of the one or more biomarkers; and (d) output means indicating that if the metric indicates an improvement or resolution of one or more symptoms of MASLD, the treatment may be discontinued or administered at a reduced dose amount and / or frequency of administration. In an embodiment, the device includes a non-transitory computer-readable medium for storing the data.

[0201] In another aspect, the disclosure features a computerized system including a data storage circuit adapted to store measurements of a biomarker level produced by the method of any one of the preceding aspects. In an embodiment, the computerized system includes a non-transitory computer-readable medium for storing the measurements.

[0202] In another aspect, the disclosure features a computer program product, including computer executable instructions for causing a computer, diagnostic arrangement, apparatus, or device to perform the steps of the method of any one of the preceding aspects. In an embodiment, the computer program product is stored within a non-transitory computer-readable medium.

[0203] In another aspect, the disclosure features a computer implemented method for detection of a MASLD disease state in a subject including: (a) receiving data on a level of expression one or more biomarkers associated with MASLD (e.g., the one or more biomarkers of Table 1 , e.g., two, three, four, five, six, seven, eight, nine, ten, or more of the biomarkers of Table 1); (b) applying the data to a predictive computer model generated based on values of one or more biomarkers in a reference population (e.g., a reference population of subjects with a MASLD disease state) or from a training set; and (c) displaying a risk probability score of the MASLD disease state in the subject on the graphical user interface. In an embodiment, the MASLD disease state is MASLD, MAFL, MASH, at-risk MASH, MASH F2, MASH F3, MASH F4, MASH F2-F3, or MASH F3-F4, with or without liver cirrhosis. In other embodiments, the subject has a NAS score >4 (e.g., a NAS score of 4, NAS score of 5, NAS score of 6, NAS score of 7, or a NAS score of 8). In other embodiments, the predictive model based on the reference population or the training set is generated from a plurality of subjects with a specific MASLD disease state. In other embodiments, the computer implemented method specifically identifies a MASLD disease state in the subject (e.g., one of MASLD, MAFL, MASH, at-risk MASH, MASH F2, MASH F3, MASH F4, MASH F2- F3, or MASH F3-F4, with or without liver cirrhosis). In yet other embodiments, the computer implemented method further identifies a treatment particularly tailored to the subject based on their identified MASLD disease state.

[0204] In another aspect, the disclosure features a computer program product, including computer executable instructions for causing a computer, diagnostic arrangement, apparatus, or device to perform the steps of the method of any of the foregoing aspects.

[0205] In another aspect, the disclosure features a computer-implemented method for classifying a human subject as having MASH F2-F3 including: (a) inputting a level of the at least one biomarker selected from the group consisting of: 3-ureidopropionate, a-ketoglutarate, and kynurenine, determined using a blood sample from a subject into a gradient-boosting machine previously trained using verified liver disease reference cases; and (b) outputting a binary classification that the subject does or does not exhibit MASH F2-F3 based on a comparison of the level of the at least one biomarker of the subject to the reference cases in the gradient boosting machine. In some embodiments, the method achieves an area- under-the-receiver-operating-characteristic curve (AUG) of at least 0.90.

[0206] In another aspect, the disclosure features a computer-implemented method for classifying a human subject as having MASH F2-F3 including: (a) obtaining a blood sample from the subject; (b) determining a level of at least one biomarker selected from the group consisting of: 3-ureidopropionate, a- ketoglutarate, and kynurenine; (c) inputting the level of the at least one biomarker into a gradient-boosting machine previously trained on biopsy-verified reference cases; and (d) outputting a binary classification that the subject does or does not exhibit MASH F2-F3. In some embodiments, the method achieves an area-under-the-receiver-operating-characteristic curve (AUG) of at least 0.90.

[0207] In another aspect, the disclosure features a computer-implemented method for classifying a human subject as having MASH F4 including: (a) inputting a level of the at least one biomarker selected from the group consisting of: 3-ureidopropionate, a-ketoglutarate, and kynurenine, determined using a blood sample from a subject into a gradient-boosting machine previously trained using verified liver disease reference cases; and (b) outputting a binary classification that the subject does or does not exhibit MASH F4 based on a comparison of the level of the at least one biomarker of the subject to the reference cases in the gradient boosting machine. In some embodiments, the method achieves an AUG of at least 0.95.

[0208] In another aspect, the disclosure features a computer-implemented method for classifying a human subject as having MASH F4 including: (a) optionally obtaining a blood sample from the subject; (b) determining a level of at least one biomarker selected from the group consisting of: 3-ureidopropionate, a- ketoglutarate, and kynurenine; (c) inputting the level of the at least one biomarker into a gradient-boosting machine previously trained on biopsy-verified reference cases; and (d) outputting a binary classification that the subject does or does not exhibit MASH F4. In some embodiments, the method achieves an AUG of at least 0.95.

[0209] In some embodiments, determining the level of at least one biomarker includes liquidchromatography tandem-mass-spectrometry (LC-MS / MS).

[0210] In some embodiments, the at least one biomarker is 3-ureidopropionate. In some embodiments, the at least one biomarker is a-ketoglutarate. In some embodiments, the at least one biomarker is kynurenine.

[0211] In some embodiments, the at least one biomarker includes at least two biomarkers or at least three biomarkers. In some embodiments: (a) the at least two biomarkers include: (I) 3-ureidopropionate and kynurenine, (ii) 3-ureidopropionate and a-ketoglutarate, or (iii) kynurenine and a-ketoglutarate; or (b) the at least three biomarkers include 3-ureidopropionate, kynurenine, and a-ketoglutarate.

[0212] In some embodiments, the method further includes inputting one or more pre-determined clinical variables into the gradient-boosting machine. In some embodiments, the one or more predetermined clinical variables include aspartate aminotransferase (AST) levels, alanine aminotransferase (ALT) levels, body mass index (BMI), the total number of components of the metabolic syndrome (MetS), platelet counts, albumin and / or age.

[0213] In some embodiments, the method further includes inputting one or more pre-determined clinical variables selected from AST levels, ALT levels, BMI, and the total number of components of the MetS. In some embodiments, the method further includes inputting one or more pre-determined clinical variables selected from the group consisting of platelet count, albumin levels, BMI, and age.

[0214] In another aspect, the disclosure features a computer program product (e.g., a software), including computer executable instructions for causing a computer, diagnostic arrangement, apparatus, or device to perform the steps of the computer implemented method of any one of the forgoing aspects.

[0215] In another aspect, the disclosure features a non-transitory computer-readable storage medium having stored thereon instructions which, when executed by one or more processors, cause the processor(s) to: (a) receive numerical inputs including: (i) at least one metabolite concentration selected from 3-ureidopropionate, a-ketoglutarate, and kynurenine, and optionally (ii) a predetermined clinical variable selected from the group consisting of: AST levels, ALT levels, BMI, the total number of components of the MetS, platelet counts, albumin levels, and / or age; (b) standardize or normalize the inputs; (c) apply an initial CatBoost gradient-boosting decision-tree model having: (i) loss_function = “Logloss”; (ii) iterations = 200; (iii) learning_rate = 0.05; (iv) depth = 6; (v) I2_leaf_reg = 1 , and / or (vi) bootstrap_type = “MVS”, random seed = 2, od_type = “Iter”, od_wait = 20; (d) generate a probability that the subject belongs to either fibrosis stage F2-F3 or stage F4 depending on the selected weight set; (e) compare the probability to a threshold that maximizes Youden's J statistic; and (f) output a binary classification of (i) “Eligible” or “Not-elig ible” for MASH F2-F3, or (ii) “Cirrhosis" or “No-cirrhosis” for MASH F4.

[0216] In some embodiments, the program is implemented in the R programming language and delivered through a Shiny graphical user interface. In some embodiments, the CatBoost model weights are containerized within a Docker image digitally signed to ensure integrity.

[0217] In some embodiments, the non-transitory computer-readable storage medium further includes instructions that, prior to step (c), automatically perform hyper-parameter tuning by exhaustive search over iterations 200-600, learning_rate 0.05-0.20, depth 5-9, and I2_leaf_reg 1 -6, selecting the combination that maximizes area-under-the-receiver-operating-characteristic curve (AUG).

[0218] In some embodiments, hyper-parameter tuning is terminated early if the validation AUC fails to improve for 20 consecutive iterations. In some embodiments, the execution time for steps (b)-(f) is less than one second on a system having four virtual CPU cores.

[0219] In another aspect, the disclosure features a computer-implemented method including: (a) deploying the non-transitory computer-readable storage medium of any of the foregoing aspects within a hospital information-technology environment; (b) receiving subject data via an electronic health record interface and simplifying its structure into a comma-separated value (CSV) file; (c) executing the Catboost model; and (d) writing a binary result back to the electronic medical record without human intervention. In some embodiments, the model weights correspond to the MASH F2-F3 classifier and achieves an AUC of at least 0.90 on unseen data. In some embodiments, the model weights correspond to the MASH F4 classifier and achieve an AUC of at least 0.95 on unseen data.

[0220] In another aspect, the disclosure features a computer-implemented method for diagnosing MASH F2-F3 in a human subject, including: (a) obtaining from the subject a numeric value for each of the following seven biomarkers: (i) ALT; (ii) TotalMetS; (iii) AST; (iv) BMI; (v) 3-ureidopropionate; (vi) a- ketoglutarate; and (vii) kynurenine; (b) assigning the value for each biomarker to one of three risk tiers (low, grey, or high) by comparing the value to two pre-determined numeric limits, LIM1 and LIM2, wherein LIM1 < LIM2 for the biomarker; (c) computing a cumulative fibrosis score (S) by adding 0 points for a low tier assignment, 1 point for a grey tier assignment and 2 points for a high tier assignment for each biomarker; and (d) classifying the subject as having MASH F2-F3 when S > 6, and as stage F0-F1 otherwise.

[0221] In some embodiments, the LIM1 and LIM2 limits for each biomarker are:

[0222] In some embodiments, a missing or non-detectable value for ALT, AST or BMI is automatically treated as low risk for this specific variable. In some embodiments, the classification in step (d) further outputs the probability (P) of MASH F2-F3 (P(F2-F3)) produced by a gradient-boosted ensemble of depth-8 symmetric decision trees trained on a reference cohort, and displays both S and P(F2-F3) to the clinician. In some embodiments, the ensemble includes 200 trees and is trained with the CatBoost algorithm. In some embodiments, the limits LIM1 and LIM2 are the two most frequently used split thresholds mined from the trained ensemble. In some embodiments, S > 8 triggers automatic referral for liver biopsy or specialist evaluation.

[0223] In another aspect, the disclosure features a non-transitory computer-readable medium storing instructions that, when executed by one or more processors, perform the method of any one of the preceding aspects.

[0224] In another aspect, the disclosure features a diagnostic system including: (a) an input interface configured to receive the biomarker values; (b) a processor configured to execute the instructions that when executed, perform the method of any one of the preceding aspect; and (c) a user interface configured to display the risk-tier assignments, cumulative score S, and P(F2-F3).

[0225] In some embodiments, 3-ureidopropionate is used as a surrogate marker of abnormal uracil catabolism, kynurenine as a surrogate of host-microbiome tryptophan metabolism, and a-ketoglutarate as a surrogate of tricarboxylic acid cycle dysregulation, thereby covering any biochemical analyte or analytical platform that captures the same metabolic pathways.

[0226] In another aspect, the disclosure features the computer-implemented method in any of the foregoing aspects for diagnosing MASH F2-F3 in a human subject, wherein the method further includes repeating steps (a)-(d) after initiation of an anti-MASH therapy and flagging a responder status when the follow-up cumulative score S2decreases by >2 points relative to baseline (Si).

[0227] In some embodiments, steps (a)-(d) are applied to predict a binary clinical outcome selected from: cirrhosis (MASH F4), cardiovascular event, hepatocellular carcinoma, liver transplantation, or allcause mortality. In some embodiments, steps (a)-(d) are applied to any disease state for which a labelled training set exists, the two-limit per variable abstraction being automatically learned from the underlying decision-tree ensemble.

[0228] In some embodiments, the method further includes providing, on a non-transitory computer- readable medium, executable source code that: (a) loads a binary CatBoost-format model file (.cbm); (b) parses the oblivious_trees object to extract all split borders for each float feature; (c) identifies, for every feature, the two borders that occur with highest frequency; (d) stores said borders as limits LIM1 and LIM2 in a human-readable table; and (e) implements steps (b)-(d) using those limits.

[0229] In some embodiments, the computer-readable medium has source code that is written in a language selected from Python, R, JavaScript, or any combination thereof, and the CatBoost model is loaded via a call to the CatBoost().load_model() API. In some embodiments, the source code automatically generates a self-contained Python or R script that hard-codes the LIM1 and LIM2 limits and the scoring logic, thereby enabling deployment of a lightweight interpretability layer without requiring the full CatBoost runtime.

[0230] In some embodiments, the computer-implemented methods for diagnosing MASH F2-F3 in a human subject in any of the foregoing aspects further includes a training pipeline which (a) ingests a labelled dataset including biomarker vectors and binary clinical outcomes, (b) trains a gradient-boosted ensemble of symmetric decision trees with depth < 8, and (c) serializes the trained ensemble into said binary CatBoost-format file.

[0231] In some embodiments, steps (a)-(c) are executed inside an automated continuous-integration workflow that triggers re-training and re-serialization whenever new outcome-labelled data are appended to the training set.

[0232] In another aspect, the disclosure features a kit including: (a) analytical reagents or calibrated sensors for quantifying at least one biomarker selected from ALT, AST, a-ketoglutarate, 3- ureidopropionate, TotalMetS components, BMI, and kynurenine; and (b) the computer-readable medium in any of the foregoing aspects, wherein the kit, when used according to the instructions, produces a cumulative fibrosis score.

[0233] In some embodiments, the computer-implemented methods for diagnosing MASH F2-F3 in a human subject in any of the foregoing aspects further includes presenting to a clinician a graphical user interface that simultaneously displays (a) the per-biomarker tier assignments, (b) the cumulative score S, (c) the model-derived P(F2-F3), and (d) a recommended clinical action selected from lifestyle intervention, pharmacologic therapy, further imaging, or liver biopsy.

[0234] In some embodiments, the same source-code framework is applied to a training dataset for a second binary outcome selected from cardiovascular event, hepatocellular carcinoma, liver-related mortality, overall mortality, response vs non-response to anti-MASH therapy, or development of type 2 diabetes, and wherein the resulting two-limit per-variable scoring table is generated and deployed.

[0235] In another aspect, the disclosure features a computer-implemented method for identifying cirrhotic fibrosis stage F4 in a human subject, including: (a) obtaining numeric values for each of the following seven biomarkers: (I) platelet count; (II) BMI; (ill) serum albumin; (iv) age; (v) 3-ureidopropionate; (vi) a-ketoglutarate; and (vii) kynurenine; (b) assigning each biomarker to one of three risk tiers — low, grey or high — by comparing the value with two limits LIM1 < LIM2 specific to that biomarker; (c) computing a cumulative cirrhosis score C by adding 0 points for a low-tier assignment, 1 point for a greytier assignment and 2 points for a high-tier assignment for every biomarker; and (d) classifying the subject as cirrhotic (F4) when C > 7, and as non-cirrhotic (F0-F3) otherwise.

[0236] In some embodiments, the tier limits are:

[0237] In some embodiments, the limits LIM1 and LIM2 are the two most frequently used split borders extracted from a gradient-boosted ensemble of depth-8 symmetric decision trees. In some embodiments, the ensemble includes 200 trees and is stored in CatBoost binary format. In some embodiments, missing platelet, BMI, or albumin values are automatically treated as low risk. In some embodiments, C > 9 triggers automatic referral for portal-hypertension work-up or transplant evaluation.

[0238] In another aspect, the disclosure features a non-transitory computer-readable medium including source code that: (a) loads a CatBoost model file; (b) parses all float-feature splits to identify the two highest-frequency borders per feature; (c) writes those borders into a human-readable table; and (d) executes the aforementioned steps (b)-(d) on patient data.

[0239] In some embodiments, the code automatically regenerates the table and redeploys the scoring script whenever the model is retrained on an updated dataset. In some embodiments, the cirrhosis-score framework is further applied to predict a binary outcome selected from decompensation, variceal bleeding, hepatocellular carcinoma, liver-transplant-free survival, or all-cause mortality (using the same two-limit abstraction learned from any CatBoost ensemble trained on corresponding labelled datasets).

[0240] In some embodiments, the biomarker-classification and scoring steps are implemented in an interpreted-language script that: (a) imports the catboost, dplyr and pROC libraries; (b) extracts a training frame and a validation frame each containing at least one metabolomic feature and the binary outcome; (c) performs hyper-parameter grid search over iterations, learning_rate, depth and I2_leaf_reg; (d) selects the parameter set that maximizes validation AUC; and (e) trains a final CatBoost model with that parameter set. In some embodiments, the grid includes 200-600 iterations, learning-rates 0.05-0.20, tree depth 5-9 and L2-leaf-regularisation 1-6.

[0241] In some embodiments, the biomarker-classification and scoring further includes: (a) evaluating the trained model on a combined dataset to obtain per-subject probabilities; (b) selecting an operating threshold that maximizes Youden’s J statistic on the ROC curve; and (c) calculating accuracy, sensitivity, specificity, positive-predictive value and negative-predictive value at that threshold.

[0242] In some embodiments, the script serializes the trained model to a CatBoost binary file named by concatenating the outcome label and selected feature names with underscore separators and the extension “.cbm.” In some embodiments, the biomarker-classification and scoring further includes storing a CSV file of the selected hyper-parameters and a plain-text printout of the rounded performance metrics. In some embodiments, the same script is re-executed automatically whenever the training dataset is modified, thereby regenerating the model file, parameter CSV and metrics without manual intervention. In some embodiments, the grid-search, model-training, threshold-selection and metric-reporting steps are parallelized across multiple CPU cores via a doParallel or foreach backend.

[0243] In another aspect, the disclosure features a diagnostic kit including (i) reagents or sensors for quantifying one or more biomarkers in any of the foregoing aspects and (ii) the computer-readable medium of any of any of the foregoing aspect, whereby the kit produces both raw probability output and thresholded class assignment in real time.

[0244] In some embodiments of any of the foregoing aspects, the method further includes a featureselection workflow that: (a) performs recursive-feature-elimination (RFE) with a CatBoost estimator to obtain per-feature importance; (b) performs principal-component analysis (PCA) on the same numeric predictors and records the Euclidean loading magnitude of each feature on the first two principal components; and (c) fits an elastic-net logistic-regression model and records the absolute value of each non-zero coefficient.

[0245] In some embodiments, the RFE step uses ten-fold cross-validation and a CatBoost base learner trained for 100 iterations with a log-loss objective. In some embodiments, the PCA step calculates the loading magnitude as ^(PC12+ PC22) for every feature. In some embodiments, the elastic-net step selects the mixing parameter a via grid search from 0.1 to 0.9 and the regularization strength A via internal cross-validation.

[0246] In some embodiments, the method of any of the foregoing aspects further includes normalizing each of the three importance measures to its own maximum, summing the normalized values to obtain a combined score, and ranking features by that combined score.

[0247] In some embodiments, the top-N ranked features (N < 20) are merged with a predetermined list of clinical covariates to form the final predictor set for model training. In some embodiments, the ranked feature list and the combined scores are exported to a CSV file titled “top_selected_features_per_combined_score.csv”.

[0248] In some embodiments, the computer-readable medium of any of the foregoing aspects further includes storing an R script that embodies the feature-selection workflow that: (a) performs recursivefeature-elimination (RFE) with a CatBoost estimator to obtain per-feature importance; (b) performs principal-component analysis (PCA) on the same numeric predictors and records the Euclidean loading magnitude of each feature on the first two principal components; and (c) fits an elastic-net logistic- regression model and records the absolute value of each non-zero coefficient, and further serializes the resulting CatBoost model to a filename derived from the outcome label concatenated with “FS.cbm”.

[0249] In another aspect, the disclosure features a kit including: (a) LC-MS / MS calibration standards for 3-ureidopropionate, kynurenine, and a-ketoglutarate; (b) reagents for protein precipitation, drying, and reconstitution; and (c) non-transitory computer-readable media storing the gradient-boosting weight sets described in any of the foregoing aspects.

[0250] BRIEF DESCRIPTION OF THE DRAWINGS

[0251] The patent or application file contains at least one drawing in color. Copies of this patent or patent application publication with color drawings will be provided by the Office upon request and payment of the necessary fee.

[0252] FIGs. 1A-1 J show an illustration showing the study design, dimensionality, and variability of metabolomic and lipidomic measurements across populations, as described further in Example 1 . FIG. 1 A represents an overview of the study design and training - validation split; FIGs. 1 B-1C represents an overview of measurements; FIGs. 1 D-1 E represents an overview of analyses; FIGs. 1 F-1G represent a PCA of the top 100 metabolites (FIG. 1 F) and 50 top lipids (FIG. 1G), showing dimensionality across NAS score stages, at-risk MASH, and country of origin; FIG. 1H is a combined partial least squares discriminant analysis (PLS-DA) of top 100 metabolites and 50 lipids for the detection of at-risk MASH, noting the complete absence of lipids; FIGs. 11-1 J are dot plots of the top metabolomic (FIG. 11) and lipidomic (FIG. 1 J) characteristics according to Benjamin! Hochberg False Discovery Rate (BH-FDR)- corrected p-values in one way ANOVAs across the Fibrosis score (left-sided) and the NAS score (rightsided).

[0253] FIGs. 2A-2O show an overview of the metabolomic, lipidomic, and clinical changes across histology in the entire cohort, as described further in Example 1 . FIGs. 2A-2C are heatmaps of the top 100 normalized metabolite concentrations with significant variability in either histological score across fibrosis (FIG. 2A) and NAS (FIG. 2B) with additional tracks indicating significant upregulation / downregulation of each metabolite in binary histopathological outcomes (FIG. 2C); FIGs. 2D-2F are heatmaps for the top 50 normalized lipid concentrations after excluding the triglyceride surge, across fibrosis (FIG. 2D) and NAS (FIG. 2E), and further indication of significant upregulation or downregulation of each lipid in binary histopathological outcomes (FIG. 2F); FIGs. 2G-2I are heatmaps for additional applicable clinical, biochemical, and hormonal measurements, across fibrosis (FIG. 2G) and NAS (FIG. 2H), and further indication of significant up or downregulation of each lipid in binary histopathological outcomes (FIG. 21); FIG. 2J shows reference graphs of composite scores (e.g., steatosis, inflammation, and fibrosis) across all heatmap columns and binary states of MASLD, MASH, and at-risk MASH; FIGs. 2K-2M show intra-cluster patterns across histological scores for fibrosis and NAS, for commonly-regulated metabolites, lipids, and clinical variables after hierarchical clustering, accompanied by pathway analysis of top metabolites; FIGs. 2N-2O show Spearman correlation coefficients of metabolites, lipids, and clinical variables with the fibrosis score (FIG. 2N) or the NAS score (FIG. 20).

[0254] FIGs. 3A-3X show plots of the complete metabolomic and lipidomic signatures under six MASLD- associated disease states (e.g., MASLD, MASH, fibrosis, fibrosis > 2, at-risk MASH, and fibrosis > 3) as compared to the absence of MASLD-associated disease states. FIG. 3A, FIG. 3E, FIG. 31, FIG. 3M, FIG. 3Q, and FIG. 3U show volcano plots of changes in the circulating metabolome under MASLD (FIG. 3A), MASH (FIG. 3E), fibrosis (FIG. 31), fibrosis > 2 (FIG. 3M), at-risk MASH (FIG. 3Q), and fibrosis > 3 (FIG. 3U), where dots represent unpaired Welch’s t-tests of metabolites. The y-axis shows the decimal logarithm of the p-value of each test, and the x-axis shows the Iog2 fold change (MASLD-associated disease state - non-MASLD-associated disease state) with positive values thus indicating metabolites increased under MASLD-associated disease state. Red and blue dots indicate metabolites with absolute fold-changes >1 .25. More opaque dots indicate variables that remain significant after the BH-FDR correction. Black dots indicate metabolites outside thresholds, and grey dots non-significant metabolites; FIG. 3B, FIG. 3F, FIG. 3J, FIG. 3N, FIG. 3R, and FIG. 3V are bar plots of the top significantly enriched metabolite pathways by FDR-significant proteins, per Fisher’s exact test under MASLD (FIG. 3B), MASH (FIG. 3F), fibrosis (FIG. 3J), fibrosis > 2 (FIG. 3N), at-risk MASH (FIG. 3R), and fibrosis > 3 (FIG. 3V). Values indicate the p-value of each Fisher’s test. Dots indicate individual metabolite fold-changes underlying overall pathway up- or downregulation; FIG. 3C, FIG. 3G, FIG. 3K, FIG. 30, FIG. 3S, and FIG. 3W are volcano plots of the changes in the circulating lipidome under MASLD (FIG. 3C), MASH (FIG. 3G), fibrosis (FIG. 3K), fibrosis > 2 (FIG. 30), at-risk MASH (FIG. 3S), and fibrosis > 3 (FIG. 3W), presented likewise, with dot colors indicating lipid classes and dot sizes indicating BH-FDR significance; FIG. 3D, FIG. 3H, FIG. 3L, FIG. 3P, FIG. 3T, and FIG. 3X represent aggregate enrichment analyses of lipid classes and secondary fatty acid saturation, length, and w3 / w6 status, based on the BH-FDR- significant lipids from the volcano plot under MASLD (FIG. 3D), MASH (FIG. 3H), fibrosis (FIG. 3L), fibrosis > 2 (FIG. 3P), at-risk MASH (FIG. 3T), and fibrosis > 3 (FIG. 3X).

[0255] FIGs. 4A-4F show the co-regulation of metabolomic, lipidomic, and clinical variables in at-risk MASH. FIG. 4A-1 and FIG. 4A-2 show a Circos plot of significant strong Spearman correlations of metabolites (absolute coefficients >0.6, after FDR correction) hierarchically clustered and analyzed for intra-cluster pathway enrichment. The outer track shows dot plots of the top significantly enriched metabolite pathways per Fisher’s exact test (p<0.05), with dot color indicating p-value (smaller = red) and dot size indicating pathway fold-enrichment. The intermediate tracks contain metabolite names and normalized levels per NAS score levels (inner tracks = lower stages, outer tracks = higher stages). The innermost track contains ribbons representing pairwise intra- and inter-cluster correlations; blue indicating negative coefficients, and red indicating positive; FIG. 4B shows hierarchical clusters of significant, very strong Spearman correlations of lipid species (absolute coefficients >0.9 after FDR correction) after hierarchical clustering. Dot size indicates the number of connections (correlations), and degree thickness indicates the strength of the coefficient (thickest = closest to 1 ). Commonly colored dots indicate coregulated families of molecules as identified by the WalkTrap method. FIG. 4C shows a hierarchically clustered Weighted Network Correlation Analysis (WCNA) of metabolites (FIG. 4C) or lipid species (FIG. 4E), and clinical features in at-risk MASH. Asterisks Indicate significant correlation eigenvalues, *, **, *** for p<0.05, 0.01 , and 0.001 , respectively; FIG. 4D is a pathway enrichment analysis of each module (1 -5) of the WCNA clusters, with dot sizes indicating fold-enrichment and color eigenvalue p-value; FIG. 4F shows intra-module lipid composition profiles. Inner circle shows parent lipid class, intermediate circle fatty acid saturation, and outer circle fatty acid length.

[0256] FIGs. 5A-5R are graphs and plots that demonstrate the building and evaluation of non-invasive machine learning models for the detection of at-risk MASH, fibrosis >3, and fibrosis >2. FIG. 5A, FIG. 5G, and FIG. 5M, PCA of variables used in the conservative model of at-risk MASH (FIG. 5A), fibrosis >3 (FIG. 5G), and fibrosis >2 (FIG. 5M), entailing all pre-determined variables and up to 2 additional omics variables per ascending order; FIG. 5B, FIG. 5H, and FIG. 5N, evaluation of the conservative Categorical Gradient Boosting Machines (CatBoost) model (green curves) within the training (10x cross-validated, 10x resampled), validation (5x cross-validated, 10x resampled) and entire (10x cross-validated, 10x resampled) cohorts for at-risk MASH (FIG. 5B), fibrosis >3 (FIG. 5H), and fibrosis >2 (FIG. 5N); FIG. 5C, FIG. 51, and FIG. 50, improvement of model performance in the validation cohort during the progression of sequential floating feature selection for at-risk MASH (FIG. 5C), fibrosis >3 (FIG. 51), and fibrosis >2 (FIG. 50); FIG. 5D, FIG. 5J, and FIG. 5P, RCA of variables used in the full model of at-risk MASH (FIG. 5D), fibrosis >3 (FIG. 5 J) , and fibrosis >2 (FIG. 5P), after sequential floating feature selection, maximizing performance with the minimal number of variables; FIG. 5E, FIG. 5K, and FIG. 5Q, evaluation of the full CatBoost model (blue curves within the training (1 Ox cross-validated, 10x resampled), validation (5x cross-validated, 10x resampled) and entire (1 Ox cross-validated, 10x resampled) cohorts for at-risk MASH (FIG. 5E), fibrosis >3 (FIG. 5K), and fibrosis >2 (FIG. 5Q); FIG. 5F, FIG. 5L, and FIG. 5R, leave-one-out cross-validation of the full model in the entire cohort for at-risk MASH (FIG. 5F), fibrosis >3 (FIG. 5L), and fibrosis >2 (FIG. 5R).

[0257] FIGs. 6A-6R are graphs and plots illustrating the building and evaluation of non-invasive machine learning models for the detection of any fibrosis, any MASH, and any MASLD. FIG. 6A, FIG. 6G, and FIG. 6M show a RCA of variables used in the conservative model of any fibrosis (FIG. 6A), any MASH (FIG. 6G), or any MASLD (FIG. 6M), entailing all pre-determined variables and up to 2 additional omics variables per ascending order; FIG. 6B, FIG. 6H, and FIG. 6N show an evaluation of the conservative CatBoost model (green curves) within the training (1 Ox cross-validated, 10x resampled), validation (5x cross-validated, 10x resampled) and entire (1 Ox cross-validated, 10x resampled) cohorts for any fibrosis (FIG. 6B), any MASH (FIG. 6H), or any MASLD (FIG. 6N); FIG. 6C, FIG. 61, and FIG. 60 show an improvement of model performance in the validation cohort during the progression of sequential floating feature selection for any fibrosis (FIG. 6C), any MASH (FIG. 61), or any MASLD (FIG. 60); FIG. 6D, FIG. 6J, and FIG. 6P show a PCA of variables used in the full model of any fibrosis (FIG. 6D), any MASH (FIG. 6J), or any MASLD (FIG. 6P), after sequential floating feature selection, maximizing performance with the minimal number of variables; FIG. 6E, FIG. 6K, and FIG. 6Q show an evaluation of the full CatBoost model (blue curves) within the training (10x cross-validated, 10x resampled), validation (5x crossvalidated, 10x resampled) and entire (10x cross-validated, 10x resampled) cohorts for any fibrosis (FIG. 6E), any MASH (FIG. 6K), or any MASLD (FIG. 6Q); FIG. 6F, FIG. 6L, and FIG. 6R represents a leave- one-out cross-validation of the full model in the entire cohort for any fibrosis (FIG. 6F), any MASH (FIG. 6L), or any MASLD (FIG. 6R).

[0258] FIGs. 7A-7B show a 5-fold cross-validated and 10x resampled evaluation of our CatBoost models and applicable non-invasive indices on the validation cohort (FIG. 7A-1 - 7A-2) and entire cohorts (FIG. 7B-1 - 7B-2) for the detection of at-risk MASH.

[0259] FIGs. 8A-8B show a complex heatmap of top 200 metabolite (FIG. 8A-1 - 8A4) and lipid (FIG. 8B-1 - 8B-4) normalized concentrations. Top features indicated by significant variability after FDR- correction in either histological score across both fibrosis and NAS with additional tracks indicating significant up / downregulation of each metabolite in binary histopathological outcomes.

[0260] FIGs. 9A-9F illustrate the co-regulation of metabolomic, lipidomic, and clinical variables in MASLD. FIG. 9A is Circos plot of significant strong Spearman correlations of metabolites (absolute coefficients >0.6, after FDR correction) hierarchically clustered and analyzed for intra-cluster pathway enrichment. The outer track shows dot plots of the top significantly enriched metabolite pathways per Fisher’s exact test (p<0.05), with dot color indicating p-value (smaller = red) and dot size indicating pathway fold-enrichment. The intermediate tracks contain metabolite names and normalized levels per NAS score levels (inner tracks = lower stages, outer tracks = higher stages). The innermost track contains ribbons representing pairwise intra- and inter-cluster correlations; blue indicating negative coefficients, and red indicating positive. FIG. 9B shows hierarchical clusters of significant very strong Spearman correlations of lipid species (absolute coefficients >0.9 after FDR correction) after hierarchical clustering. Dot size indicates the number of connections (correlations), and degree thickness indicates the strength of the coefficient (thickest = closest to 1 ). Commonly colored dots indicate co-regulated families of molecules as identified by the WalkTrap method. FIG. 9C shows hierarchically clustered WCNA of metabolites and clinical features in at-risk MASH. Asterisks Indicate significant correlation eigenvalues, *, **, *** for p<0.05, 0.01 and 0.001 respectively. FIG. 9D shows a pathway enrichment analysis of each module (1 -5) of the WCNA clusters, with dot sizes indicating fold-enrichment and color eigenvalue p-value. FIG. 9E is a WCNA of lipid species and clinical features likewise presented. FIG. 9F illustrates intra-module lipid composition profiles. Inner circle shows parent lipid class, intermediate circle fatty acid saturation, and outer circle fatty acid length.

[0261] FIGs. 10A-10F illustrate the co-regulation of metabolomic, lipidomic, and clinical variables in MASH. FIG. 10A is a circos plot of significant strong Spearman correlations of metabolites (absolute coefficients >0.6, after FDR correction) hierarchically clustered and analyzed for intra-cluster pathway enrichment. The outer track shows dot plots of the top significantly enriched metabolite pathways per Fisher’s exact test (p<0.05), with dot color indicating p-value (smaller = red) and dot size indicating pathway fold-enrichment. The intermediate tracks contain metabolite names and normalized levels per NAS score levels (inner tracks = lower stages, outer tracks = higher stages). The innermost track contains ribbons representing pairwise intra- and inter-cluster correlations; blue indicating negative coefficients, and red indicating positive. FIG. 10B is a hierarchical clusters of significant very strong Spearman correlations of lipid species (absolute coefficients >0.9 after FDR correction) after hierarchical clustering. Dot size indicates the number of connections (correlations), and degree thickness indicates the strength of the coefficient (thickest = closest to 1 ). Commonly colored dots indicate co-regulated families of molecules as identified by the WalkTrap method. FIG. 10C is a hierarchically clustered WCNA of metabolites and clinical features in at-risk MASH. Asterisks Indicate significant correlation eigenvalues, *, **, *** for p<0.05, 0.01 and 0.001 respectively. FIG. 10D is a pathway enrichment analysis of each module (1 -5) of the WCNA clusters, with dot sizes indicating fold-enrichment and color eigenvalue p- value. FIG. 10E shows a WCNA of lipid species and clinical features likewise presented. FIG. 10F illustrates intra-module lipid composition profiles. Inner circle shows parent lipid class, intermediate circle fatty acid saturation, and outer circle fatty acid length.

[0262] FIGs. 11A-11 F illustrate the co-regulation of metabolomic, lipidomic, and clinical variables in fibrosis. FIG. 11 A is a circos plot of significant strong Spearman correlations of metabolites (absolute coefficients >0.6, after FDR correction) hierarchically clustered and analyzed for intra-cluster pathway enrichment. The outer track shows dot plots of the top significantly enriched metabolite pathways per Fisher’s exact test (p<0.05), with dot color indicating p-value (smaller = red) and dot size indicating pathway fold-enrichment. The intermediate tracks contain metabolite names and normalized levels per NAS score levels (inner tracks = lower stages, outer tracks = higher stages). The innermost track contains ribbons representing pairwise intra- and inter-cluster correlations; blue indicating negative coefficients, and red indicating positive. FIG. 11 B shows hierarchical clusters of significant very strong Spearman correlations of lipid species (absolute coefficients >0.9 after FDR correction) after hierarchical clustering. Dot size indicates the number of connections (correlations), and degree thickness indicates the strength of the coefficient (thickest = closest to 1 ). Commonly colored dots indicate co-regulated families of molecules as identified by the WalkTrap method. FIG. 11C is a hierarchically clustered WCNA of metabolites and clinical features in at-risk MASH. Asterisks Indicate significant correlation eigenvalues, *, ", *** for p<0.05, 0.01 and 0.001 respectively. FIG. 11D is a pathway enrichment analysis of each module (1 -5) of the WCNA clusters, with dot sizes indicating fold-enrichment and color eigenvalue p- value. FIG. 11 E is a WCNA of lipid species and clinical features likewise presented; FIG. 11 F shows intramodule lipid composition profiles. Inner circle shows parent lipid class, intermediate circle fatty acid saturation, and outer circle fatty acid length.

[0263] FIGs. 12A-12F illustrate the co-regulation of metabolomic, lipidomic, and clinical variables in significant fibrosis. FIG. 12A is a circos plot of significant strong Spearman correlations of metabolites (absolute coefficients >0.6, after FDR correction) hierarchically clustered and analyzed for intra-cluster pathway enrichment. The outer track shows dot plots of the top significantly enriched metabolite pathways per Fisher’s exact test (p<0.05), with dot color indicating p-value (smaller = red) and dot size indicating pathway fold-enrichment. The intermediate tracks contain metabolite names and normalized levels per NAS score levels (inner tracks = lower stages, outer tracks = higher stages). The innermost track contains ribbons representing pairwise intra- and inter-cluster correlations; blue indicating negative coefficients, and red indicating positive. FIG. 12B shows hierarchical clusters of significant very strong Spearman correlations of lipid species (absolute coefficients >0.9 after FDR correction) after hierarchical clustering. Dot size indicates the number of connections (correlations), and degree thickness indicates the strength of the coefficient (thickest = closest to 1 ). Commonly colored dots indicate co-regulated families of molecules as identified by the WalkTrap method. FIG. 12C shows a WCNA of metabolites and clinical features in at-risk MASH. Asterisks Indicate significant correlation eigenvalues, *, **, *" for p<0.05, 0.01 and 0.001 respectively. FIG. 12D is a pathway enrichment analysis of each module (1 -5) of the WCNA clusters, with dot sizes indicating fold-enrichment and color eigenvalue p-value. FIG. 12E is a WCNA of lipid species and clinical features likewise presented. FIG. 12F shows intra-module lipid composition profiles. Inner circle shows parent lipid class, intermediate circle fatty acid saturation, and outer circle fatty acid length.

[0264] FIGs. 13A-13S illustrate a comprehensive evaluation of the machine learning models for at-risk MASH. FIGs. 13A-13C show ROC curves of conservative model performance in the training, validation, and entire cohorts, respectively, annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUG, accuracy, PPV and NPV. FIGs. 13D-13F show ROC curves of floating model performance in the training, validation, and entire cohorts, respectively, annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUG, accuracy, PPV and NPV. FIG. 13G is a graph of inter-individual Shapley values of each variable’s contribution in model architecture. FIG. 13H shows learning curves plotting aggregate AUROC in slices of the training (blue) and validation (red) data, resampled 100 times. FIGs. 13I-13J show subgroup evaluation of the floating model in participants with (FIG. 131) and without (FIG. 13J) diabetes mellitu, annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUC, accuracy, PPV and NPV; likewise for participants with (FIG. 13K) and without (FIG. 13L) obesity; ikewise for participants from gastroenterology / hepatology (FIG. 13M) or bariatric (FIG. 13N) clinics; likewise for participants with MASLD (FIG. 130), and excluding participants with cirrhosis (FIG. 13KP); and likewise for participants from Australia (FIG. 13Q), Italy (FIG. 13R), and Greece (FIG. 13S).

[0265] FIGs. 14A-14B illustrate a comparison of machine learning models for at-risk MASH against 15 indices in the entire cohort. FIG. 14A are violin plots of 5-fold cross-validated, 10x resampled evaluation of models and demonstration of metrics as individual datapoints across violin plots. CatBoost for the floating model, and CB conservative for the cojnservative model. Machine learning models represented by red violins, whereas steatosis-specific models with yellow, MASH-specific models with green, and fibrosis-specific models with blue. The table shows p-values of the unpaired Welch’s t-tests comparing the AUROCs of all models with either floating or conservative catboost models, and optimal threshold indicating the mean Youden’s optimal threshold across all folds. FIG. 14B shows a non-cross-validated evaluation of model performance across the cohort, compared per Delong’s test. Metrics displayed with 95% confidence intervals with the exception of F1 (ratio). Regression for a simple logistic regression model implementing a conventional random forest recursive feature elimination; Catboost optimized for a version of the floating model with optimized hyperparameters without nested cross-validation, but rather based on the traits of the overall cohort in question. Delong’s vs for the p-values of the Delong’s tests of applicable models against each catboost or regression model.

[0266] FIGs. 15A-15B illustrate a comparison of machine learning models for at-risk MASH against 15 indices in the validation cohort. FIG. 15A are violin plots of 5-fold cross-validated, 10x resampled evaluation of models and demonstration of metrics as individual datapoints across violin plots. CatBoost for the floating model, and CB conservative for the conservative model. Machine learning models represented by red violins, whereas steatosis-specific models with yellow, MASH-specific models with green, and fibrosis-specific models with blue. The table shows p-values of the unpaired Welch’s t-tests comparing the AUROCs of all models with either floating or conservative catboost models, and optimal threshold indicating the mean Youden’s optimal threshold across all folds. FIG. 15B shows a non-cross- validated evaluation of model performance across the cohort, compared per Delong’s test. Metrics displayed with 95% confidence intervals with the exception of F1 (ratio). Regression for a simple logistic regression model implementing a conventional random forest recursive feature elimination; Catboost optimized for a version of the floating model with optimized hyperparameters without nested cross- validation, but rather based on the traits of the overall cohort in question. Delong’s vs for the p-values of the Delong’s tests of applicable models against each catboost or regression model.

[0267] FIGs. 16A-16B illustrate a comparison of machine learning models for at-risk MASH against 15 indices in a random data split. FIG. 16A are violin plots of 5-fold cross-validated, 10x resampled evaluation of models and demonstration of metrics as individual datapoints across violin plots. CatBoost for the floating model, and CB conservative for the conservative model. Machine learning models represented by redf violins, whereas steatosis-specific models with yellow, MASH-specific models with green, and fibrosis-specific models with blue. Tables represent p-values of the unpaired Welch’s t-tests comparing the AUROCs of all models with either floating or conservative catboost models, and optimal threshold indicating the mean Youden’s optimal threshold across all folds. FIG. 16B shows a non-cross- validated evaluation of model performance across the cohort, compared per Delong's test. Metrics displayed with 95% confidence intervals with the exception of F1 (ratio). Regression for a simple logistic regression model implementing a conventional random forest recursive feature elimination; Catboost optimized for a version of the floating model with optimized hyperparameters without nested cross- validation, but rather based on the traits of the overall cohort in question. Delong’s vs for the p-values of the Delong's tests of applicable models against each catboost or regression model.

[0268] FIG. 17 is an illustration showing the study design, dimensionality, and variability of metabolomic and lipidomic measurements across populations, as described further in Example 2.

[0269] FIGs. 18A-18I are graphs demonstrating a cross-validated and resampled model evaluation across the training, validation, and entire cohorts. Green, blue and red curves represent the ROC curves of the conservative (FIG. 18A-18C), floating (FIG. 18D-18F), and optimized models (FIG. 18G-18I). FIG. 18A, FIG. 18D, and FIG. 18G are 10-fold cross-validated, 10x resampled evaluation of the respective models within the training cohort; FIG. 18B, FIG. 18E, and FIG. 18H are 5-fold cross-validated, 10x resampled evaluation of the respective models within the validation cohort; and FIG. 18C, FIG. 18F, and FIG. 181 are 10-fold cross-validated, 10x resampled evaluation of the respective models in the entire cohort. Within each graph, the mean AUG, mean accuracy, mean sensitivity, and mean specificity, along with the confidence intervals, represents the mean area under the curve and accuracy of all iterations, and sensitivity-specificity the mean Youden’s optimal sensitivity-sensitivity pairs of all iterations. ALT, alanine aminotransferase; AST, aspartate aminotransferase; AUG, Area under the receiver operating characteristics curve; BMI, body mass index; MetS, Metabolic syndrome.

[0270] FIGs. 19A-19C are graphs showing the leave-one-out cross-validation of model performances within the entire cohort. Blue curves represent sensitivity, red curves specificity, and green dotted lines the optimal ML probability threshold (negative = ruling out of the condition). Intra-figure tables represent model metrics at the optimal Youden’s index. AUC, Area under the receiver operating characteristics curve; NPV, Negative predictive value; PPV, Positive predictive value.

[0271] FIGs. 20A-20E are graphs illustrating subgroup ROC curves of the floating model according to diabetes status (FIG. 20A), obesity (FIG. 20B), histologically-confirmed MASLD (FIG. 20C), clinic type (FIG. 20D) and country (FIG. 20E). Red dots represent the optimal Youden’s index at each curve alongside the corresponding sensitivity and specificity. Blue annotations represent the AUG, accuracy, PPV and NPV at each respective curve. AUC, Area under the receiver operating characteristics curve; NPV, Negative predictive value; PPV, Positive predictive value.

[0272] FIGs. 21A-21 E are graphs illustrating subgroup ROC curves of the optimized model according to diabetes status (FIG. 21A), obesity (FIG. 21B), histologically-confirmed MASLD (FIG. 21C), clinic type (FIG. 21 D) and country (FIG. 21 E). Red dots represent the optimal Youden’s index at each curve alongside the corresponding sensitivity and specificity. Blue annotations represent the AUC, accuracy, PPV and NPV at each respective curve. AUC, Area under the receiver operating characteristics curve; NPV, Negative predictive value; PPV, Positive predictive value.

[0273] FIGs. 22A-22D show cross-validated and resampled evaluation of the model across the validation and entire cohorts. FIG. 22A and FIG. 22C are violin plots and individual data points that represent each metric across every iteration. Squares indicate the mean metric, with error bars indicating the standard deviation. Red squares represent the CatBoost models, yellow squares steatosis-specific models, green squares MASH-specific models, and blue squares fibrosis-specific models. FIG. 22B and FIG. 22D are tables detailing Welch’s unpaired t-tests comparing all aggregate AUCs with those of the CatBoost models, with significance threshold set at p=10-5. Optimal threshold indicating the model thresholds for all NITs, and Youden’s index for the CatBoost models. AUG, Area under the receiver operating characteristics curve; NPV, Negative predictive value; PPV, Positive predictive value.

[0274] FIGs. 23A-23E show an analysis of co-regulation of metabolomic, lipidomic, and clinical variables in fibrosis > 3. FIG. 23A shows a Circos plot of significant strong Spearman correlations of metabolites (absolute coefficients >0.6, after FDR correction) hierarchically clustered and analyzed for intra-cluster pathway enrichment. The outer track shows dot plots of the top significantly enriched metabolite pathways per Fisher’s exact test (p<0.05), with dot color indicating p-value (smaller = red) and dot size indicating pathway fold-enrichment. The intermediate tracks contain metabolite names and normalized levels per NAS score levels (inner tracks = lower stages, outer tracks = higher stages). The innermost track contains ribbons representing pairwise intra- and inter-cluster correlations; blue indicating negative coefficients, and red indicating positive; FIG. 23B shows hierarchical clusters of significant very strong Spearman correlations of lipid species (absolute coefficients >0.9 after FDR correction). Dot size indicates the number of connections (correlations), and degree thickness indicates the strength of the coefficient (thickest = closest to 1 ). Commonly colored dots indicate co-regulated families of molecules as identified by the WalkTrap method. FIG. 23C shows hierarchically clustered Weighted Network Correlation Analysis (WCNA) of metabolites and clinical features in at-risk MASH. Asterisks Indicate significant correlation eigenvalues, *, **, *** for p<0.05, 0.01 and 0.001 respectively; FIG. 23D shows a pathway enrichment analysis of each module (1 -5) of the WCNA clusters, with dot sizes indicating fold-enrichment and color eigenvalue p-value; FIG. 23E shows intra-module lipid composition profiles. Inner circle shows parent lipid class, intermediate circle fatty acid saturation, and outer circle fatty acid length.

[0275] FIGs. 24A-24R show a comprehensive evaluation of the machine learning models for fibrosis > 3. FIG. 24A-24C are area under the curve (AUROC) graphs of conservative model performance in the training, validation, and entire cohorts, respectively, annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUG, accuracy, PPV and NPV. FIG. 24D-24F are AUROC graphs of floating model performance in the training, validation, and entire cohorts, respectively, annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUG, accuracy, PPV and NPV. FIG. 24G, graph of inter-individual Shapley values of each variable’s contribution in model architecture; FIG. 24H shows learning curves plotting aggregate AUROC in slices of the training (blue) and validation (red) data, resampled 100 times; FIG. 241- 24J are subgroup evaluation of the floating model in participants with diabetes mellitus (FIG. 241) and without diabetes mellitus (FIG. 24J), annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUC, accuracy, PPV and NPV; FIG. 24K-24L is a subgroup evaluation of the floating model in participants with obesity (FIG. 24K) and without obesity (FIG. 24L), annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUC, accuracy, PPV and NPV; FIG. 24M-24N, subgroup evaluation of the floating model in participants gastroenterology / hepatology (FIG. 24M) or bariatric (FIG. 24N) clinics, annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUC, accuracy, PPV and NPV; FIG. 240 is a subgroup evaluation of the floating model in participants with MASLD, annotated by the Youden’s optimal index (red dot), corresponding sensitivity and specificity levels, as well as overall AUC, accuracy, PPV and NPV; FIG. 24P-24R shows subgroup evaluation of the floating model in participants from Australia (FIG. 24P), Italy (FIG. 24Q), and Greece (FIG. 24R).

[0276] FIGs. 26A-26D show graphs of cross-validated and 10x resampled model evaluation for MASH F2-F3 detection. ROC curves represent basic (FIG. 26A), basic + grid search (FIG. 26B), floating (FIG. 26C), and floating + grid search (FIG. 26D) models. Mean AUC, accuracy, sensitivity, and specificity with confidence intervals are reported. AUC, Area under the ROC curve; BMI, Body mass index.

[0277] FIGs. 27A-27D show graphs of cross-validated and 10x resampled model evaluation for cirrhosis (MASH F4) detection. ROC curves represent basic (FIG. 27A), basic + grid search (FIG. 27B), floating (FIG. 27C), and floating + grid search (FIG. 27D) models. Mean AUC, accuracy, sensitivity, and specificity with confidence intervals are reported. AUC, Area under the ROC curve; BMI, Body mass index.

[0278] FIGs. 28A-28D show graphs of leave-one-out cross-validation of the MASH F2-F3 and cirrhosis models within the validation and entire cohorts. Basic (FIG. 28A), Basic+grid search (FIG. 28B), floating (FIG. 28C), and floating+grid search (FIG. 28D) are shown, with curves indicating sensitivity and specificity, and dotted lines indicating the optimal machine learning probability cutoff. Overall metrics are noted beside each graph. ALT, alanine aminotransferase; AST, aspartate aminotransferase; AUC, area under the curve; BMI, body mass index; GDF-15, Growth differentiation factor 15; IGF-1 , Insulin-like growth factor 1 .

[0279] FIGs. 29A-29H show Inter-individual Shapley values per variable / observation across all models in patients with MASH F2-F3 (FIG. 29A, FIG. 29C, FIG. 29E, and FIG. 29G) or MASH F4 (FIG. 29B, FIG. 29D, FIG. 29F, and FIG. 29H)

[0280] FIGs. 30A-30B are graphs showing aggregate ROC curves in the validation cohort showcasing changes in model performance during the floating sequential feature selection process in patients with MASH F2-F3 (FIG. 30A) or MASH F4 (“cirrhosis”; FIG. 30B).

[0281] FIGs. 31 A-31 H are graphs showing the secondary sensitivity analysis described in Example 7 that cross-validated and resampled model evaluations across the training (bariatric, Italy), validation (Gl, Australia + Greece), and entire cohorts. Curves represent the ROC curves for detection of cirrhosis of the basic (FIG. 31A-31B), basic + grid search (FIG. 31C-31D), floating (FIG. 31-E-31 F), and floating + grid search (FIG. 31G-31 H) models, respectively. FIG. 31 A, FIG. 31C, FIG. 31 E, FIG. 31G show the 10-fold cross-validated, 10x resampled evaluation of the respective models within the training cohort; FIG. 31B, FIG. 31 D, FIG. 31 F, FIG. 31 H show the 5-fold cross-validated, 10x resampled evaluation of the respective models within the validation (gastroenterology) cohort. Within-graph mean AUC, accuracy, sensitivity and specificity, alongside their respective confidence intervals, represent the mean area under the curve and accuracy of all iterations, and sensitivity-specificity the mean Youden’s optimal sensitivitysensitivity pairs of all iterations. ALT, alanine aminotransferase; AST, aspartate aminotransferase; AUC, Area under the receiver operating characteristic curve; BMI, body mass index; MetS, Metabolic syndrome.

[0282] DEFINITIONS

[0283] To facilitate an understanding of the disclosure, a number of terms are defined below. Terms defined herein have meanings as commonly understood by a person of ordinary skill in the areas relevant to the invention. Terms such as "a", "an," and "the" are not intended to refer to only a singular entity, but include the general class of which a specific example may be used for illustration. The terminology herein is used to describe specific embodiments of the disclosure, but their usage does not limit the disclosure, except as outlined in the claims.

[0284] As used herein, the term "about," as applied to one or more values of interest, refers to a value that falls within 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1%, or less in either direction (greater than or less than) of a stated reference value, unless otherwise stated or otherwise evident from the context (except where such number would exceed 100% of a possible value).

[0285] As used herein, any values provided in a range of values include both the upper and lower bounds, and any values contained within the upper and lower bounds.

[0286] As used herein, “administration" refers to providing or giving a subject a therapy (e.g., a therapeutic agent), such as, e.g., a therapy for MASH, at-risk MASH, or liver fibrosis. When the therapy is a therapeutic agent, the therapeutic agent can be administered by any effective route for that agent. Exemplary routes of administration are oral administration, parenteral injection, intraperitoneal injection, intrathecal injection, intraventricular injection, intraarticular injection, intravenous injection, subcutaneous injection, intramuscular injection, intranasal or inhalation, or topically.

[0287] As used herein, a “control” is any useful reference (e.g., a sample, such as a reference sample) used to diagnose MASLD, MAFL, MASH, at-risk MASH, or liver fibrosis. The control can be any sample, standard, standard curve, or level (e.g., mean or median) that is used for comparison purposes. The control may be a negative control (e.g., a sample or level from a subject diagnosed as not having MASLD, MAFL, MASH, at-risk MASH, or liver fibrosis, e.g., a healthy subject) or a positive control (e.g., a sample or level from a subject clinically diagnosed as having MASLD, MAFL, MASH, at-risk MASH, or liver fibrosis).

[0288] As used herein a “reference level” or “reference standard” is a value or number derived from a sample (e.g., a reference sample), such as a level of one or more biomarkers (e.g., 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and / or TAG 54 :4 / FA22:4) in a biological sample (e.g., a blood, serum, or plasma sample) obtained from a healthy subject (e.g., a negative control subject, such as a subject that does not have MASLD, MAFL, MASH, at-risk MASH, or liver fibrosis). The reference level can be based on measurements from a single subject or from a population of subjects (i.e., a reference population) and may represent raw data or data transformed using machine learning algorithms. If a reference level is derived from a reference population (e.g., a population of subjects that does not have MASLD, MAFL, MASH, at-risk MASH, or liver fibrosis), the reference level may be an expression level at the 50thpercentile of the reference population, or the 60thpercentile, or the 70thpercentile, or the 80thpercentile, or the 90thpercentile, or greater. The reference level is used for diagnosis of a subject as having MASLD, MAFL, MASH, at-risk MASH, or liver fibrosis. The reference level may be obtained from a reference sample that is matched to the sample obtained from the subject by at least one (e.g., at least one, at least two, at least three, at least four, or at least five or more) of the following criteria: age, weight, sex, disease stage, and overall health. For example, the negative control sample for diagnosis of an obese subject (e.g., a subject having a BMI greater than or equal to than 27.5 kg / m2) may be a serum sample from an obese subject (e.g., a subject having a BMI greater than or equal to than 27.5 kg / m2) not having MASLD, MAFL, MASH, at-risk MASH, or liver fibrosis.

[0289] As used herein, the terms “relative to a reference level” refers to a comparison between a subject’s biomarker level (e.g., a level of 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and / or TAG54:4 / FA22:4 in a sample (e.g., a blood sample) obtained from the subject or a level of 3- ureidopropionate, kynurenine, and / or a-ketoglutarate in a sample (e.g., a blood sample) obtained from the subject) and a reference (e.g., a control) level of the same biomarker. For example, the following formula can be used to compare the level of a biomarker disclosed herein in a sample (e.g., a blood sample) from a subject (e.g., a level of 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and / or

[0290] TAG54:4 / FA22:4) relative to a control level of the same biomarker.

[0291] Subject's Biomarker Level

[0292] Fold Change =

[0293] Control Level

[0294] As used herein, the term “increase” in the context of a level of a biomarker refers to an increase of 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more in the level of the biomarker relative to a control level.

[0295] As used herein, the term “decrease” in the context of a level of a biomarker disclosed herein refers to a decrease of 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, or more in the level of said biomarker relative to a control level.

[0296] As used herein a “reference sample” is a sample from a normal healthy subject (e.g., a negative control), such as a blood sample (e.g., a serum or plasma sample) from a subject not having MASLD, MAFL, MASH, at-risk MASH (e.g., MASH F2-F3 or MASH F4), or liver fibrosis.

[0297] By “diagnosing” is meant identifying a molecular or pathological state, disease, or condition, such as the identification of MALSD, MASH, MAFL, at-risk MASH (e.g., MASH F2-F3 or MASH F4), or liver fibrosis or to refer to identification of a subject having MASLD, MASH, MAFL, at-risk MASH (e.g., MASH F2-F3 or MASH F4), or liver fibrosis who may benefit from a particular treatment regimen or therapy.

[0298] As used herein, the terms “effective amount,” “therapeutically effective amount,” and a “sufficient amount” of a composition or therapy described herein refer to a quantity sufficient to, when administered to the subject in need thereof, including a mammal, for example a human, effect beneficial or desired results, including clinical results, and, as such, an “effective amount” or synonym thereto depends upon the context in which it is being applied. For example, in the context of treating MASH, at-risk MASH (e.g., MASH F2-F3 or MASH F4), or liver fibrosis, it is an amount of the composition or therapy sufficient to achieve a treatment response as compared to the response obtained without administration of the composition of therapy. The amount of a given composition described herein that will correspond to such an amount will vary depending upon various factors, such as the given agent, the pharmaceutical formulation, the route of administration, the type of disease or disorder, the identity of the subject (e.g. age, sex, weight) being treated, and the like, but can nevertheless be routinely determined by one skilled in the art. Also, as used herein, a “therapeutically effective amount” of a composition or therapy of the present disclosure is an amount which results in a beneficial or desired result in a subject as compared to a control. Note that when a combination of active ingredients is administered, the effective amount of the combination may or may not include amounts of each ingredient that would have been effective if administered individually. As defined herein, a therapeutically effective amount of a composition or therapy of the present disclosure may be readily determined by one of ordinary skill by routine methods known in the art. A dosage regimen may be adjusted to provide the optimum therapeutic response.

[0299] As used herein, the term “MASLD therapy” and “liver fibrosis therapy” refers to a therapy that is recognized as effective for treating MASLD (e.g., MAFL, MASH, or at-risk MASH, such as MASH F2-F3 or MASH F4) and fibrosis, respectively. For example, a MASLD therapy can be, or may comprise, weight loss, a dietary change, increased physical activity, an ani-fibrotic agent, an anti-inflammatory agent, an anti-apoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a peroxisome prol iterator-activated receptor (PPAR) agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a farnesoid X receptor (FXR) agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), JKB- 121 , resmetirom (REZDIFFRA™), a liver transplant. For example, a fibrosis therapy can be weight loss, a dietary change, increased physical activity, an ani-fibrotic agent, an anti-inflammatory agent, an antiapoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a peroxisome prol iterator-activated receptor (PPAR) agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a farnesoid X receptor (FXR) agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), JKB- 121 , resmetirom (REZDIFFRA™), or a liver transplant. Other MASLD and / or fibrosis therapies may include an immunotherapy, a chemotherapy, a radiation therapy, an ablation therapy, a targeted drug therapy, a surgical-based therapy, or any combination thereof. Exemplary immunotherapies include immune checkpoint inhibitors, such as a programmed cell death 1 (PD-1 ) inhibitor (e.g., pembrolizumab (KEYTRUDA®) or nivolumab (e.g., OPDIVO®), a programmed death-ligand 1 (PD-L1 ) inhibitor (e.g., atezolizumab (e.g., TECENTRIQ®) or durvalumab (e.g., IMFINZI®)), or a cytotoxic T-lymphocyte associated protein 4 (CTLA-4) inhibitor (e.g., ipilimumab (e.g., YERVOY®) or tremelimumab (e.g., IMJUDO®)). Exemplary chemotherapies include gemcitabine (e.g., GEMZAR®), oxaliplatin (e.g., ELOXATIN®), cisplatin, doxorubicin (pegylated liposomal doxorubicin) (e.g., ADRIAMYCIN®), 5- fluorouracil (5-FU), capecitabine (e.g., XELODA®), mitoxantrone (e.g., NOVANTRONE®) and gemcitabine plus oxaliplatin (e.g., GEMOX®). Exemplary radiation therapies include external beam radiation therapy (EBRT), stereotactic body radiation therapy (SBRT), image guided radiation therapy (IGRT), 3D conformal radiation therapy (3DCRT), intensity modulated radiation therapy (IMRT), volumetric modulated arc therapy (VMAT), and radioembolization therapy. Exemplary ablation therapies include radiofrequency ablation (RFA), microwave ablation (MWA), cryoablation (cryotherapy), and ethanol (alcohol) ablation. Exemplary targeted drug therapies include kinase inhibitors (e.g., sorafenib (NEXAVAR®), lenvatinib (e.g., LENVIMA®), regorafenib (e.g., STIVARGA®), and cabozantinib (e.g., CABOMETYX®)) and monoclonal antibodies (e.g., bevacizumab (e.g., AVASTIN®) and ramucirumab (e.g., CYRAMZA®). Exemplary surgical-based therapies include partial hepatectomy and liver transplants. Other liver therapies include, but are not limited to, empagliflozin (e.g., JARDIANCE®), evogliptin, exenatide (e.g., BYETTA®), febuxostat (e.g., ULORIC®), gliclazide (e.g., DIAMICRON®), glimepiride (e.g., AMARYL®), ipragliflozin (e.g., SUGLAT®), liraglutide (e.g., VICTOZA®), lobeglitazone, metformin (e.g., GLUCOPHAGE®), pentoxifylline, rifampicin (e.g., RIFATER®), salsalate, sitagliptin (e.g., JANUVIA®), tofogliflozin, a T cell immunoreceptor with immunoglobulin and ITIM (TIGIT) inhibitor (e.g., vibostolimab, etigilimab, tiragolumab, domvanalimab, M6223, ociperlimab, and EOS884448), and a lymphocyte-activation gene 3 (LAG3) targeting immunotherapy (e.g., relatlimab, eftilagimod alpha, favezelimab, fianlimab, tebotelimab, and RO7247669).

[0300] As used herein, the terms “metabolic dysfunction-associated steatotic liver disease” and “MASLD”, also known in the art as “nonalcoholic fatty liver disease” and “NAFLD”, refer to a condition in which fat (e.g., lipids) build up in the liver due to causes other than excessive alcohol consumption. Metabolic dysfunction-associated liver (MAFL) and metabolic dysfunction-associated steatohepatitis (MASH) are two types of MASLD. During the pathogenesis of MASLD, certain conditions of the liver may arise, such as steatosis (e.g., the presence of lipid droplets in hepatocytes) and fibrosis (e.g., the scarring of the liver tissue) during earlier stages of MASLD, and at-risk MASH (e.g., MASH F2-F3 or MASH F4), cirrhosis (e.g., liver scarring and damage, e.g., stage 4 fibrosis (F4 fibrosis)), and liver cancer (e.g., hepatocellular carcinoma) during the mid-to-later stages of MASLD.

[0301] As used herein, the terms “metabolic dysfunction-associated fatty liver” and “MAFL”, also known in the art as “nonalcoholic fatty liver” and “NAFL” refer to a condition in which the liver contains excess fat but does not exhibit liver inflammation or liver cell damage.

[0302] As used herein, the terms “metabolic dysfunction-associated steatohepatitis” and “MASH”, also known in the art as “non-alcoholic associated steatohepatitis” and “NASH”, refer to an advanced form of MASLD in which a buildup of fat in the liver leads to liver inflammation and liver cell damage.

[0303] As used herein, the term “at-risk MASH” refers to MASH that is accompanied by a fibrosis score of > 2 (e.g., F2, F2-F3, F3, or F4) and, optionally, a non-alcoholic steatohepatitis activity (NAS) score >4. “MASH F2,” “MASH F2-F3,” “MASH F3,” and “MASH F3-F4” are subtypes of at-risk MASH.

[0304] As used herein, the term “MASH F2-F3” refers to at risk MASH with a fibrosis score of 2 (F2) or a fibrosis score of about 3 (F3). As used herein, the term “MASH F3-F4” refers to at risk MASH with a fibrosis score of 3 (F3) or a fibrosis score of 4 (F4). As used herein, the term “MASH F4” refers to at risk MASH with a fibrosis score of 4 (F4).

[0305] As used herein, the term “liver fibrosis” refers to the buildup of scar tissue in the liver, which may be caused by repetitive or long-lasting injury or inflammation.

[0306] As used herein, the term “sample” refers to a specimen (e.g., blood, blood component (e.g., serum or plasma), urine, saliva, amniotic fluid, cerebrospinal fluid, tissue (e.g., placental or dermal), pancreatic fluid, chorionic villus sample, and cells) isolated from a subject. As used herein, the terms “subject” and “patient” refer to an animal (e.g., a mammal, such as a human), veterinary animals (e.g., cats, dogs, cows, horses, sheep, pigs, etc.) and experimental animal models of diseases (e.g., mice, rats). A subject to be diagnosed and treated according to the methods described herein may be one who has been diagnosed with MASL, MAFL, MASH, at-risk MASH, or liver fibrosis. Diagnosis may be performed using the methods described herein. One skilled in the art will understand that a subject to be treated according to the present disclosure may have been subjected to standard tests or may have been identified, without examination, as one at risk due to the presence of one or more risk factors associated with the disease or condition (e.g., obesity).

[0307] As used herein, “treatment” and “treating” in reference to a disease or condition, refer to an approach for obtaining beneficial or desired results, e.g., clinical results. Beneficial or desired results can include alleviation or amelioration of one or more symptoms or conditions; diminishment of extent of disease or condition; stabilized (i.e., not worsening) state of disease, disorder, or condition; preventing spread of disease or condition; delay or slowing the progress of the disease or condition; amelioration or palliation of the disease or condition; and remission (whether partial or total), whether detectable or undetectable. “Ameliorating” or “palliating” a disease or condition means that the extent and / or undesirable clinical manifestations of the disease, disorder, or condition are lessened and / or time course of the progression is slowed or lengthened, as compared to the extent or time course in the absence of treatment. “Treatment” can also mean prolonging survival as compared to expected survival if not receiving treatment. Those in need of treatment include those already with the condition or disorder, as well as those prone to have the condition or disorder or those in which the condition or disorder is to be treated prophylactical ly (e.g., prior to the development of one or more symptoms or overt symptom presentation).

[0308] DETAILED DESCRIPTION

[0309] The present disclosure features biomarkers and methods of using the biomarkers to differentiate between a healthy subject and a subject with or at risk of developing metabolic dysfunction-associated steatotic liver disease MASLD (e.g., metabolic dysfunction-associated fatty liver (MAFL) or metabolic dysfunction-associated steatohepatitis (MASH), e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH with a fibrosis score of F2 to F3 (“MASH F2-F3”), MASH with a fibrosis score of F3 to F4 (“MASH F3-F4”), or MASH with a fibrosis score of F4 (“MASH F4”).

[0310] MASLD affects more than 2 billion people worldwide including over 100 million Americans, representing a trillion-dollar 10-year healthcare market estimate. MASH F2-F3, for example, represents a critical and reversible stage of MASLD necessitating methods of early intervention. Existing non-invasive tests (NIT; e.g., legacy serum scores and elastography-based imaging) were never designed or analytically validated for MASH F2-F3. Moreover, their diagnostic accuracy collapses whenever cirrhosis (F4) must be excluded or, conversely, specifically confirmed. Additionally, imaging modalities, such as vibration-controlled transient elastography, magnetic-resonance elastography, and ultrasound stiffness panels are capital-intensive, largely unavailable in primary-care clinics, and demonstrably inaccurate for detecting MASH itself, let alone for discriminating the critical F2-F3 fibrosis stage. Remarkably, the biomarkers and methods described herein provide a highly accurate and sensitive NIT for MASH F2-F3, MASH F3-F4, and MASH F4.

[0311] Exemplary biomarkers are provided in Tables 1A-D along with their relative expression during MASLD progression (either an increase or a decrease in a level of the biomarkers observed in a MASLD subject relative to a control subject). Accordingly, the methods described herein can be used to diagnose MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject, monitor treatment efficacy (e.g., efficacy of a therapy for MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4)) in a subject receiving the treatment, and to monitor disease development and / or progression in a subject with MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4)). These methods provide a simple and non-invasive alternative to liver biopsy and provide a highly-sensitive and accurate alternative to conventional non-invasive tests (NITs).

[0312] Table 1A. Exemplary Biomarkers Upregulated in MASLD

[0313]

[0314] Table 1B. Exemplary Biomarkers Downregulated in MASLD

[0315] Table 1C. Exemplary Biomarkers Upregulated in At-risk MASH Table 1D. Exemplary Biomarkers Downregulated in At-risk MASH

[0316] MASLD and Fibrosis

[0317] MASLD, the hepatic component of metabolic syndrome, is now recognized as the most prevalent chronic liver disease worldwide, affecting 25-30% of the worldwide population. In the first stage of MASLD, an accumulation of fat is observed in the liver (MAFL), which can progress to an inflammatory state (MASH) and, in later stages, to significant liver fibrosis (at-risk MASH) and cirrhosis (typically stage F4 liver fibrosis) that can lead to hepatocellular carcinoma (HCC). Liver fibrosis has the following stages during disease progression: stage 0 (F0), which indicates no fibrosis; stage 1 (F1), which indicates centrilobular pericellular fibrosis (F1); stage 2 (F2), which indicates centrilobular and periportal fibrosis; stage 3 (F3), which indicates bridging fibrosis; and stage 4 (F4) which indicates cirrhosis. Patients with MASH accompanied by significant fibrosis (F>2) are referred to as at-risk MASH (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) and have a substantial increase in morbidity and mortality. The median time to develop advanced fibrosis when inflammation is absent on the initial biopsy (patients with MAFL) is 13.4 years, while the median time when it is present (patients with MASH) is only 4.2 years.

[0318] The gold standard for the diagnosis of MASLD (e.g., MAFL, MASH, at-risk MASH, or liver fibrosis) and for assessing liver fibrosis is liver biopsy. However, liver biopsy is a costly method that carries certain risks for the patient, such as pain after biopsy, bleeding, and infection, and the liver biopsy specimen is not always representative of the actual status of the entire liver, often resulting in misclassification due to sampling error and / or interobserver viability. Imaging techniques have emerged as attractive alternatives, however, such tools are often not available in primary care or in smaller community-based gastroenterological or endocrinological departments. Additionally, there are no reliable biomarkers or scores for the non-invasive diagnosis of MASH or for staging liver fibrosis with a similar score as in liver histology. Due to the cost and invasiveness of the procedure, most patients with MASLD go undiagnosed.

[0319] The following methods are useful for addressing these issues, which is quite timely in view of the recent approval by the U.S. Food and Drug Administration (FDA) of resmetirom (e.g., REZDIFFRA®), a thyroid receptor beta agonist for the treatment of MASH F2-F3. Notably, the FDA has defined the MASH F2-F3 window as the eligibility criterion for resmetirom (and for other ongoing phase-ill trials), for which, prior to the present disclosure, there is no accurate or reliable non-invasive test (NIT) built or validated for this niche population. While legacy scores and imaging panels exist, these tests fail to achieve acceptable accuracy, even for the broader F2-F4 range which includes cirrhosis.

[0320] Methods of Diagnosing MASLD and / or Liver Fibrosis

[0321] Described herein are biomarkers that can be used alone or in combination for the diagnosis of MASLD (e.g., MAFL or MASH, e.g, at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject. These biomarkers include various lipids, glycans, fatty acids, hormones, and metabolites and can be detected in a blood sample (e.g., a serum sample or plasma sample). Diagnosis can be performed by comparing biomarker concentrations in the biological sample (e.g., a blood, serum, or plasma sample) obtained from the subject to biomarker concentrations in a control (e.g., a negative control) sample of the same type (e.g., a blood, serum, or plasma sample) or to biomarker concentrations in a biological sample previously obtained from the same subject, e.g., prior to their development of MASLD / liver fibrosis (e.g., prior to an early (e.g., F1 or F2, e.g., F1 -F2) stage of the disease), or prior to progression of their MASLD to an intermediate (e.g., F2 or F3, e.g., MASH F2-F3) or a later stage of the disease (e.g., MASH F3 or MASH F4, e.g., MASH F3-F4). The biological sample from the subject may be compared directly to a reference level, which may be based on a control sample (e.g., a negative control sample from a subject not having MASLD), a control sample collected from the subject at an earlier time point when the subject was deemed healthy, a sample collected from the subject at an earlier time point when the subject was at an earlier or less severe disease stage, or a predetermined reference value (e.g., a measurement from a single subject or from a population of subjects).

[0322] Diagnosis using biomarkers that are increased in MASLD

[0323] Diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject may be achieved by determining a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject, wherein an increased level (e.g., an increase of 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) of the at least one biomarker relative to a reference level is indicative of the presence of MASLD. Any biomarker described herein (e.g., a biomarker listed in Table 1A or Table 1 C) may be used for the diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4).

[0324] For example, the at least one biomarker may be selected from the group including: 1 -stearoyl- GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3- indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158 (or a combination thereof). Alternatively, the at least one biomarker may be selected from the group including: 3-ureidopropionate, kynurenine, a- ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N- acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine (or a combination thereof). In a preferred embodiment, the at least one biomarker may be selected from the group including: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4 (or a combination thereof). In another embodiment, the at least one biomarker may be selected from the group including: 3-ureidopropionate, kynurenine, and a-ketoglutarate, (or a combination thereof). Exemplary combinations include: 3-ureidopropionate and kynurenine; 3-ureidopropionate and a-ketoglutarate; kynurenine and a- ketoglutarate; or 3-ureidopropionate, kynurenine, and a-ketoglutarate. In some embodiments, the biomarkers are compared to reference values. In some embodiments the reference value for 3-ureidopropionate is between about -1 .2 to 2.5 arbitrary units (AU) (about -1 .2 to -1 , about -1 to -0.9 (e.g.; -0.99, -0.98, -0.97, -0.96, -0.85, -0.94, -0.93, -0.92, -0.91 ,-0.9), about -0.9 to -0.8, about -0.8 to -0.7, about -0.7 to -0.6, about -0.6 to -0.4, about -0.4 to -0.2, about -0.2 to 0, about 0 to 0.5, about 0.5 to 1 , about 1 to 1 .5, about 1 .5 to 2 (e.g.; 1 .5, 1 .6, 1 .7, 1 .8, 1 .9, 2), about 2 to 2.5 AU). In some embodiments, the reference value for kynurenine is between about -2 to 1 .5 (AU) (about -2 to -1 .5 (e.g.; -2, -1 .9, -1 .8, - 1 .7, -1 .69, -1 .65, -1 .6,- 1 .5), about -1 .5 to -1 , about -1 to -0.5, about -0.5 to 0, about 0 to 0.5, about 0.5 to 1 , about 1 to 1 .5 (e.g.; 1 to 1 .1 , 1 .1 to 1 .2, 1 .2 to 1 .3, 1 .3 to 1 .4, 1 .4 to 1 .5) AU). In some embodiments, the reference value for a-ketoglutarate is between -2 to 2.5 (AU) (about -2 to -1 .9 (e.g. -1 .99, -1 .98, -1 .97, -1 .96, -1 .95, -1 .94, -1 .93, -1 .92, -1 .91 ), about -1 .9 to -1 .8 about -1 .8 to -1 , about -1 to -0.5, about -0.5 to 0, about 0 to 0.5, about 0.5 to 1 , about 1 to 1 .5, about 1 .5 to 2, about 2 to 2.5 (e.g.; 2.1 , 2.2, 2.3, 2.4, 2.5) AU).

[0325] Combinations of at least two (e.g., at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, or at least ten, e.g., two, three, four, five, six, seven, eight, nine, or ten) biomarkers may be used for the diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). The at least two biomarkers may be selected from the group including: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-i ndoleglyoxylic acid, 3-ureidopropionate, 4- hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 - phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2- FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158. Alternatively, the at least two biomarkers may be selected from the group including: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O-methyluridine. In a preferred embodiment, the at least two biomarkers may be selected from the group including: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4. In another embodiment, the at least two biomarkers may be selected from the group including: 3-ureidopropionate, kynurenine, and a- ketoglutarate, (or a combination thereof). Exemplary combinations include: 3-ureidopropionate and kynurenine; 3-ureidopropionate and a-ketoglutarate; or kynurenine and a-ketoglutarate.

[0326] In an example, at least two biomarkers are used for the diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4), wherein the at least two biomarkers include: (i) 3-ureidopropionate and kynurenine; (II) 3-ureidopropionate and a- ketoglutarate; (iii) 3-ureidopropionate and mannonate; (iv) 3-ureidopropionate and TAG54:4 / FA22:4; (v) kynurenine and a-ketoglutarate; (vi) kynurenine and mannonate; (vii) kynurenine and TAG54:4 / FA22:4; (viii) a-ketoglutarate and mannonate; (ix) a-ketoglutarate and TAG54:4 / FA22:4; or (x) mannonate and TAG54:4 / FA22:4. In another example, at least three biomarkers are used for the diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4), wherein the at least three biomarkers include: (i) 3-ureidopropionate, kynurenine, and a-ketoglutarate; (ii) 3-ureidopropionate, kynurenine, and mannonate; (iii) 3-ureidopropionate, kynurenine, and TAG54:4 / FA22:4; (iv) 3-ureidopropionate, a-ketoglutarate, and mannonate; (v) 3-ureidopropionate, a- ketoglutarate, and TAG54:4 / FA22:4; (vi) 3-ureidopropionate, mannonate, and TAG54:4 / FA22:4; (vii) kynurenine, a-ketoglutarate, and mannonate; (viii) kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4; (ix) kynurenine, mannonate, and TAG54:4 / FA22:4; or (x) a-ketoglutarate, mannonate, and TAG54:4 / FA22:4. In another embodiment, the at least three biomarkers may be 3-ureidopropionate, kynurenine, and a- ketoglutarate.

[0327] In yet another example, at least four biomarkers are used for the diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4), wherein the at least four biomarkers include: (i) kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4; (ii) 3-ureidopropionate, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4; (iii) 3- ureidopropionate, kynurenine, mannonate, and TAG54:4 / FA22:4; (iv) 3-ureidopropionate, kynurenine, a- ketoglutarate, and TAG54:4 / FA22:4; or (v) 3-ureidopropionate, kynurenine, a-ketoglutarate, and mannonate.

[0328] In yet another example, at least five biomarkers are used for the diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4), wherein the at least five biomarkers include: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0329] The biomarkers can be detected and / or measured using any appropriate clinical assay. Biomarkers (e.g., 3-ureidopropionate, kynurenine, and / or a-ketoglutarate) can be identified by methods including, but not limited to liquid chromatography (LC), mass spectrometry (MS), gas chromatography (GO), LC-MS, liquid-chromatography tandem-mass-spectrometry (LC-MS / MS), GC-MS, glycoblotting, radioimmunoassay, ultraviolet light (UV), infrared spectroscopy, fluorescence, nuclear magnetic resonance (NMR), and enzyme-linked immunosorbent assay (ELISA). Appropriate methods of detection may depend on the molecular type (e.g., nucleic acid, protein, lipid, etc.) of the biomarker being measured and are readily apparent to one of skill in the art.

[0330] For example, mass spectrometry (e.g., LC-MS) can be used to detect and measure any biomarker described herein, such as those described in FIG. 8A, FIG. 8B, or any of the following: 3-UPA, kynurenine, a-ketoglutarate, mannonate, X-12007, vanillic acid glycine, trans-urocanate, 9, 10-DiHOME, 12, 13-DiHOME, threonylphenylalanine, gamma-glutamylcitrulline, HWESASXX, 2-deoxyuridine, fibrinopeptide B (1 -13), phenylalanineglycine, fibrinopeptide A (2-15), fibrinopeptide A (3-15), fibrinopeptide A (3-16), fibrinopeptide A (5-16), propyl 4-hydroxybenzoate, leucylalanine, N- acetylmethionine, androsterone sulfate, epiandrosterone sulfate, inosine, guanosine, dehydroepiandrosterone sulfate, pregnanediol sulfate, 1 ,5-anhydroglucitol, pregnenolone sulfate, etiocholanolone glucuronide, 2-oxoarginine, 2-keto3-deoxygluconate, 3-hydroxyisobutyrate, propionylcarnitine, succinylcarnitine, N-acetylcitrulline, 2-aminoheptanoate, N-acetylanine, aspartate, glucose, ribonate, 4-hydroxyhippurate, fructosyllysine, 2-methylcitrate, sphinganine, phenylalanine, perfluorohexanesulfonate, choline, TAG(54:4), TAG(54:4) / FA22:4), AcCa(10:0), AcCa(10:1 ), Cer(d34:0), Cer(d34:2), Cer(d18:2_25:1 ), Cer(d43:0), Cer(d43:3), DG(34:1 ), DG(34:2), DG(36:2), DG(36:4), Hex1 Cer(d34:2), LPC(20:0e), LPC(20:3), LPC(22:5), LPE(16:0), LPE(18:0), LPE(22:5), PA(44:4), PC(32:0), PC(32:1 e), PC(32:1e)=> PC(16:0e_16:1), PC(32:1 e)=> PC(16:1 e_16:0), PC(34:0), PC(34:1), PC(34:1 e), PC(34:2e), PC(35:2), PC(35:3), PC(36:2), PC(36:3), PC(36:4), PC(36:5e), PC(37:2), PC(37:3), PC(40:5e), PC(40:6e), PC(40:7), PC(40:8), PC(42:6), PE(34:1 ), PE(34:3e), PE(38:1 ), PE(38:6), PE(40:4e), Pl(36:1 ), SM(d32:0), SM(d32:2), SM(d35:1 ), SM(d36:0), SM(d36:4), SM(d37:1), SM(d38:0), SM(d38:1 ), SM(d40:1 ), 00(36:1), TG(38:0), TG(38:2), TG(40:0), TG(43:1), TG(52:4), TG(53:5), DG(36:3), LPC(18:0), PC(36:2), PC(37:2), PC(40:5), TG(38:0), TG(50:0), TG(51 :1 ), TG(57:1 ), and TG(60:2). Note: the aforementioned lipids are named based on the stereospecific numbering (sn) method, written as “Headgroup(sn1 / sn2)” in which the structures of the side chains are indicated within parentheses. Common headgroup abbreviations include PC for phosphatidylcholines (LPC for lyso species), PE for phosphatidylethanolamines (LPE for lyso species), PG for phosphatidylglycerols (LPG for lyso species), PA for phosphatidic acids (LPA for lyso species), PI for glycerophosphoinositols (LPI for lyso species), AcCa for acylcarnitines, Co for coenzyme Q10, PPA for glyceropyrophosphates, DG for diglycerides, TAG for triacylglycerol, TG for triradylglycerolipids, Cer for ceramides, and SM for sphingomyelins.

[0331] Additional biomarkers may further include those described in U.S. International Patent Application No. WO 2021 / 092265, which is incorporated herein by reference in its entirety. Furthermore, the diagnostic methods described herein can be used individually or in combination with another diagnostic method, such as imaging (e.g., abdominal ultrasound, computerized tomography scanning, magnetic resonance imaging, transient elastography, or magnetic resonance elastography) and / or liver biopsy.

[0332] Diagnosis using biomarkers that are decreased in MASLD

[0333] Diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) can be achieved by determining a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject, wherein a decreased level (e.g., a decrease of 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) of the at least one biomarker relative to a reference level is indicative of the presence of MASLD. Any biomarker described herein (e.g., a biomarker listed in Table 1B or Table 1 D) may be used for the diagnosis of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4).

[0334] For example, the at least one biomarker may be selected from the group including: androsterone glucuronide, CER(14:0), guanosine, PC(14:0 / 20:4), PC(15:0 / 18:2), pregnanediol-3-glucuronide, TAG57:3- FA18:2, 2'-deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.

[0335] Methods of Treating MASLD and / or Liver Fibrosis

[0336] Provided herein are methods of treating a subject that has been identified as in need of treatment of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3- F4, or MASH F4). The subject in need of treatment may have been previously diagnosed (e.g., by a clinician) as having MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) using methods known in the art (e.g., a subject diagnosed by ultrasound, magnetic resonance imaging (MRI), and / or magnetic resonance elastography (MRE)). Additionally or alternatively, the subjects may have been previously diagnosed as having MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) using the methods described above.

[0337] A subject can be treated by measuring a level of one or more biomarkers described herein (e.g., see Tables 1A-1D) in a biological sample (e.g., a blood, serum, or plasma sample) obtained from the subject, determining that the level of the one or more biomarkers identifies the subject as one in need of treatment of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4), and treating the subject with an appropriate therapy (e.g., a MASLD therapy or a liver fibrosis therapy). A subject may also be treated for MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) based on a prior diagnosis, such as a prior diagnosis using the methods described above or those conventional in the field of MASLD.

[0338] Treating MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2- F3, MASH F3-F4, or MASH F4) can be achieved by determining that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from a subject is increased (e.g., an increase of 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%, 125%, 150%, 175%, 200%, 300% or more) relative to a reference level and administering a therapy (e.g., a MASLD therapy or a liver fibrosis therapy) to the subject. The methods of treatment may further include a step of monitoring treatment efficacy of the MASLD therapy or liver fibrosis therapy administered to the subject, the methods of which are described herein.

[0339] Any biomarker(s) described herein (e.g., one or more biomarkers in Tables 1A-1 D) may be used to establish or to identify a subject in need of treatment of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). For example, the at least one biomarker may be selected from the group including: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158 (or a combination thereof). Alternatively, the at least one biomarker may be selected from the group including: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2'- deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O- methyluridine (or a combination thereof). In a preferred embodiment, the at least one biomarker may be selected from the group including: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4 (or a combination thereof). Combinations of at least two (e.g., at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, or at least ten, e.g., two, three, four, five, six, seven, eight, nine, or ten) biomarkers described herein may be used in the methods of treating described herein. In some embodiments, the at least one biomarker may be selected from the group including: 3-ureidopropionate, kynurenine, and a-ketoglutarate, (or a combination thereof). Exemplary combinations include: 3-ureidopropionate and kynurenine; 3-ureidopropionate and a- ketoglutarate; kynurenine and a-ketoglutarate; or 3-ureidopropionate, kynurenine, and a-ketoglutarate. Other exemplary biomarker combinations are described in the diagnostic methods section above.

[0340] Additional biomarkers for use in the methods of treatment described herein may include those described in U.S. International Patent Application No. WO 2021 / 092265, which is incorporated herein by reference in its entirety.

[0341] MASLD and liver fibrosis therapies

[0342] MASLD (e.g., MAFL, MASH, e.g., at-risk MASH, e.g., MASH F2-F3 or MASH F3-F4) therapies include or involve weight loss, dietary changes (e.g., eating a diet rich in fruits, vegetables, whole grains, and healthy fats (e.g., low glycemic index foods); reducing sugar intake (e.g., intake of simple sugars); limiting intake of fats (e.g., saturated and / or trans fats); or reducing daily calorie intake), increased physical activity, anti-inflammatory and anti-apoptosis agents (e.g., pentoxifylline, selonsertib, (GS-4997) tipelukast (MN-001 ), and emricasan), medications to treat insulin resistance or type 2 diabetes (e.g., metformin DPP4 inhibitors, sulfonyl urea, glucagon-like peptide 1 receptor agonists (e.g., liraglutide, dulagl utide, and semaglutide), and SGLT2 inhibitors (e.g., canagliflozin, luseogliflozin, ipragliflozin, and empagliflozin)), medications to improve cholesterol (e.g., ezetimibe, statins), Vitamin E, peroxisome proliferator-activated receptor (PPAR) agonists, including PPARa agonists bezafibrate, fenofibrate, and pemafibrate (K-877), PPARy agonists such as pioglitazone, INT131 , and MK0533, PPAR5 agonists, such as seladepar (MBX-8025) and endurobol (GW501516), dual PPARa / 5 agonists such as elafibranor (GFT505), dual PPARa / y agonists, such as saroglitazar, and pan-PPAR agonists such as IVA337, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, anti-hypertensive drugs (e.g., losartan), farnesoid X receptor (FXR) agonists (e.g., Obeticholic acid (OCA), INT-767, GS-9674, LMB763, LJN452), MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), and JKB- 121 . Therapies for at-risk MASH include or involve weight loss, dietary changes (e.g., eating a diet rich in fruits, vegetables, whole grains, and healthy fats (e.g., low glycemic index foods); reducing sugar intake (e.g., intake of simple sugars); limiting intake of fats (e.g., saturated and / or trans fats); or reducing daily calorie intake), increased physical activity, anti-inflammatory and anti-apoptosis agents (e.g., pentoxifylline, selonsertib, (GS-4997) tipelukast (MN-001 ), and emricasan), medications to treat insulin resistance or type 2 diabetes (e.g., metformin DPP4 inhibitors, sulfonyl urea, glucagon-like peptide 1 receptor agonists (e.g., liraglutide, dulaglutide, and semaglutide), and SGLT2 inhibitors (e.g., canagliflozin, luseogliflozin, ipragliflozin, and empagliflozin)), medications to improve cholesterol (e.g., ezetimibe, statins), Vitamin E, peroxisome proliferator-activated receptor (PPAR) agonists, including PPARa agonists bezafibrate, fenofibrate, and pemafibrate (K-877), PPARy agonists such as pioglitazone, INT131 , and MK0533, PPAR6 agonists, such as seladepar (MBX-8025) and endurobol (GW501516), dual PPARa / 6 agonists such as elafibranor (GFT505), dual PPARa / y agonists, such as saroglitazar, and pan-PPAR agonists such as IVA337, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, anti-hypertensive drugs (e.g., losartan), farnesoid X receptor (FXR) agonists (e.g., Obeticholic acid (OCA), INT-767, GS-9674, LMB763, LJN452), MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS- 986036, Cenicriviroc (CVC), JKB-121 , resmetirom (REZDIFFRA™), and liver transplant. Other exemplary therapies for MASLD (e.g., MASH, e.g., at-risk MASH, e.g., MASH F2-F3 or MASH F3-F4) therapies are discussed in Sumida and Yoneda, J. Gastroenterol. 53:362-376, 2018, which is incorporated herein by reference.

[0343] Exemplary liver fibrosis therapies include or involve weight loss, dietary changes, increased physical activity, anti-inflammatory and anti-apoptosis agents (e.g., pentoxifylline, selonsertib, (GS-4997) tipelukast (MN-001 ), and emricasan), medications to treat insulin resistance or type 2 diabetes (e.g., metformin DPP4 inhibitors, sulfonyl urea, glucagon-like peptide 1 receptor agonists (e.g., liraglutide, dulagl utide, and semaglutide), and SGLT2 inhibitors (e.g., canagliflozin, luseogliflozin, ipragliflozin, and empagliflozin)), medications to improve cholesterol (e.g., ezetimibe, statins), Vitamin E, peroxisome proliferator-activated receptor (PPAR) agonists, including PPARa agonists bezafibrate, fenofibrate, and pemafibrate (K-877), PPARy agonists such as pioglitazone, INT131 , and MK0533, PPAR5 agonists, such as seladepar (MBX-8025) and endurobol (GW501516), dual PPARa / 6 agonists such as elafibranor (GFT505), dual PPARa / y agonists, such as saroglitazar, and pan-PPAR agonists such as IVA337, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, anti-hypertensive drugs (e.g., losartan), farnesoid X receptor (FXR) agonists (e.g., Obeticholic acid (OCA), INT-767, GS-9674, LMB763, LJN452), MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, JKB-121 , resmetirom (REZDIFFRA™), anti-fibrotic agents (e.g., Cenicriviroc (CVC), Simtuzumab, GR-MD-02, and ND-LO2- s0201 ), and liver transplant if the fibrosis is severe or has progressed to cirrhosis. Other exemplary therapies for liver fibrosis are discussed in Sumida and Yoneda, J. Gastroenterol. 53:362-376, 2018, which is incorporated herein by reference. Alternatively, the subject may be treated with an experimental therapy. For example, based on the level of one or more biomarkers described herein (e.g., metabolites, lipids, glycans, hormones, and / or fatty acids), the subject may be identified as in need of treatment for any MASLD (e.g., MAFL, MASH, at-risk MASH, or liver fibrosis), and enrolled in a clinical trial for treatment with a new therapeutic for MAFL, MASH, at-risk MASH, or liver fibrosis.

[0344] Methods of Monitoring

[0345] The methods described herein can also be used to evaluate treatment efficacy and / or disease development and / or progression in a subject having MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). One or more of the biomarkers described herein (e.g., one or more biomarkers of Tables 1A-1 D) can be measured in a biological sample (e.g., a blood sample, such as a serum or plasma sample) obtained from a subject undergoing treatment and compared to a reference level, a control sample, such as a negative control sample (e.g., a sample from a subject not having MAFL, MASH, at-risk MASH, or liver fibrosis), or a biological sample obtained from the subject at an earlier timepoint (e.g., 1 week earlier, two weeks earlier, one month earlier, 2 months earlier, 3 months earlier, 4 months earlier, 6 months earlier, 9 months earlier, 1 year earlier, or more) or prior to treatment initiation. Periodically monitoring (e.g., daily, weekly, monthly, or yearly) the directionality of the one or more biomarkers described herein (e.g., one or more biomarkers of Tables 1 A-1D) in the subject can aid in a clinician’s evaluation of the subject’s disease status. In general, if a subject’s biomarker level(s) continues to increase (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) over time (e.g., over a 3 month, 6 month, 12 month, 18 month, or 24 month period), it is likely that the subject’s disease is progressing and / or that a given treatment is ineffective. Alternatively, if a subject’s biomarker level(s) decrease (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) over time (e.g., over a 3 month 6, month, 12 month, 18 month, or 24 month period), it is likely that the subject's disease is in remission and / or that a given treatment is effective. No change (e.g., <5% increase or <5% decrease) in a subject’s biomarker level(s) over time (e.g., over a 3 month, 6, month, 12 month, 18 month, or 24 month period), likely indicates that the subject’s disease status is unchanged.

[0346] Monitoring a subject’s treatment may include the step of identifying the subject’s condition as progressive or stabilized, and modifying a given therapy (e.g., a MASLD therapy or a liver fibrosis therapy) being administered to the subject. The condition may be considered progressive if it is determined that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers described herein, e.g., see Tables 1A-1D in a biological sample (e.g., blood) obtained from the subject has increased (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) over time (e.g., over a 3 month, 6 month, 12 month, 18 month, or 24 month period) relative to a reference level or a previous sample obtained from the subject. The condition may be considered stabilized if it is determined that a level of at least one biomarker in a biological sample obtained from the subject has remained unchanged over time (e.g., over a 3 month, 6 month, 12 month, 18 month, or 24 month period) relative to a reference level or a previous sample obtained from the subject.

[0347] Monitoring a subject’s treatment may include the step of identifying the subject’s condition as in remission, and modifying a given therapy (e.g., a MASLD therapy or a liver fibrosis therapy) being administered to the subject. The condition may be considered progressive if it is determined that a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers described herein, e.g., see Tables 1A-1 D) in a biological sample (e.g., blood) obtained from the subject has increased (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) over time (e.g., over a 3 month, 6 month, 12 month, 18 month, or 24 month period) relative to a reference level or a previous sample obtained from the subject. The condition may be considered stabilized if it is determined that a level of at least one biomarker in a biological sample obtained from the subject has remained unchanged (e.g., <5% increase or <5% decrease) over time (e.g., over a 3 month, 6 month, 12 month, 18 month, or 24 month period) relative to a reference level or a previous sample obtained from the subject.

[0348] If the method of monitoring indicates that the subject’s condition is stabilized or is progressive (e.g., progressed from MAFL to MASH, from MASH to at-risk MASH, from F1 to F2 liver fibrosis, from F2 to F3 liver fibrosis, or from F3 to F4 liver fibrosis), the subject’s treatment can be modified; for example, the therapeutic can be administered to the subject more frequently and / or at a higher dosage, or the subject can be treated using a different therapeutic. If the results indicate that the subject’s condition has improved or gone into remission, the subject can continue treatment with the same therapeutic at the same dose and / or frequency of administration, the frequency of administration and / or dosage of the therapeutic can be decreased, or the therapy may be discontinued. The subject may be evaluated once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or more frequently to evaluate treatment efficacy. These methods can be used to evaluate the efficacy of standard treatments or to evaluate the efficacy of experimental treatments, such as therapeutics tested in a clinical trial.

[0349] Monitoring treatment efficacy in a subject being treated with a therapy (e.g., a MASLD therapy or a liver fibrosis therapy) may include detecting an increase (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) in a level of at least one biomarker (e.g., at least one biomarker in Tables 1A-1 D) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject and modifying the MASLD therapy administered to the subject (e.g., increasing a dosage and / or frequency of administration of a therapeutic agent to the subject). The method of monitoring treatment efficacy may be performed once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, bi-monthly, monthly, bi-weekly, weekly, or daily.

[0350] Monitoring treatment efficacy in a subject being treated with a therapy (e.g., a MASLD therapy or a liver fibrosis therapy) may include detecting a decrease (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) in a level of at least one biomarker (e.g., at least one biomarker in Table 1) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject and modifying the MASLD therapy administered to the subject (e.g., increasing a dosage and / or frequency of administration of a therapeutic agent to the subject). The method of monitoring treatment efficacy may be performed once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, bi-monthly, monthly, bi-weekly, weekly, or daily.

[0351] Monitoring progression of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject being treated with a therapy (e.g., a MASLD therapy or a liver fibrosis therapy) may include detecting an increase (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) in a level of at least one biomarker (e.g., at least one biomarker in Tables 1A-1 D) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject and modifying the MASLD therapy administered to the subject, e.g., by increasing a dosage and / or frequency of administration of a therapeutic agent to the subject. The method of monitoring the progression of MASLD may be performed once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, bi-monthly, monthly, bi-weekly, weekly, or daily.

[0352] Monitoring progression of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject being treated with a therapy (e.g., a MASLD therapy or a liver fibrosis therapy) may include detecting a decrease (e.g., by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more) in a level of at least one biomarker (e.g., at least one biomarker in Tables 1A-1 D) in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject and modifying the MASLD therapy administered to the subject, e.g., by decreasing a dosage and / or frequency of administration of a therapeutic agent to the subject. The method of monitoring the progression of MASLD may be performed once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, bi-monthly, monthly, bi-weekly, weekly, or daily.

[0353] The at least one biomarker may be any biomarker described herein (e.g., see Tables 1A-1 D), such as a biomarker selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158 (or a combination thereof). In some embodiments, the at least one biomarker is selected from 3- ureidopropionate, kynurenine, and a-ketoglutarate.

[0354] Combinations of at least two (e.g., at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, or at least ten, e.g., two, three, four, five, six, seven, eight, nine, or ten) biomarkers may be used for the methods of monitoring described above. The at least two biomarkers may be selected from the group including: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O- methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4- hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 - phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2- FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158. In some embodiments, the at least two biomarkers are selected from 3-ureidopropionate, kynurenine, and / or a-ketoglutarate. For example, the at least two biomarker include: 3-ureidopropionate and kynurenine; 3-ureidopropionate and a- ketoglutarate; or kynurenine and a-ketoglutarate.

[0355] Alternatively, the at least two biomarkers may be selected from the group including: 3- ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X-26158, ALTUIL, 2’- deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2-methylcitrate / homocitrate, SAH, and 2'-O- methyluridine. In a preferred embodiment, the at least two biomarkers may be selected from the group including: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4. In an embodiment, the at least two biomarkers may be selected from the group including: 3-ureidopropionate, kynurenine, and a-ketoglutarate, (or a combination thereof). Exemplary combinations include: 3- ureidopropionate and kynurenine; 3-ureidopropionate and a-ketoglutarate; or kynurenine and a- ketoglutarate. In an example, at least two biomarkers are used for monitoring, wherein the at least two biomarkers include: (i) 3-ureidopropionate and kynurenine; (ii) 3-ureidopropionate and a-ketoglutarate; (iii) 3-ureidopropionate and mannonate; (iv) 3-ureidopropionate and TAG54:4 / FA22:4; (v) kynurenine and a- ketoglutarate; (vi) kynurenine and mannonate; (vii) kynurenine and TAG54:4 / FA22:4; (viii) a-ketoglutarate and mannonate; (lx) a-ketoglutarate and TAG54:4 / FA22:4; or (x) mannonate and TAG54:4 / FA22:4.

[0356] In another example, at least three biomarkers are used for monitoring, wherein the at least three biomarkers include: (i) 3-ureidopropionate, kynurenine, and a-ketoglutarate; (ii) 3-ureidopropionate, kynurenine, and mannonate; (iii) 3-ureidopropionate, kynurenine, and TAG54:4 / FA22:4; (iv) 3- ureidopropionate, a-ketoglutarate, and mannonate; (v) 3-ureidopropionate, a-ketoglutarate, and TAG54:4 / FA22:4; (vi) 3-ureidopropionate, mannonate, and TAG54:4 / FA22:4; (vii) kynurenine, a- ketoglutarate, and mannonate; (viii) kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4; (ix) kynurenine, mannonate, and TAG54:4 / FA22:4; or (x) a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0357] In yet another example, at least four biomarkers are used for monitoring, wherein the at least four biomarkers include: (i) kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4; (ii) 3- ureidopropionate, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4; (iii) 3-ureidopropionate, kynurenine, mannonate, and TAG54:4 / FA22:4; (iv) 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4; or (v) 3-ureidopropionate, kynurenine, a-ketoglutarate, and mannonate.

[0358] In yet another example, at least five biomarkers are used for monitoring, wherein the at least five biomarkers include: 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

[0359] Additional biomarkers for use in the methods of monitoring described herein may include those described in U.S. International Patent Application No. WO 2021 / 092265, which is incorporated herein by reference in its entirety.

[0360] The biomarkers can be detected and / or measured using any appropriate clinical assay. Biomarkers (e.g., 3-ureidopropionate, kynurenine, and / or a-ketoglutarate) can be identified by methods including, but not limited to liquid chromatography (LC), mass spectrometry (MS), gas chromatography (GO), LC-MS, liquid-chromatography tandem-mass-spectrometry (LC-MS / MS), GC-MS, glycoblotting, radioimmunoassay, ultraviolet light (UV), infrared spectroscopy, fluorescence, nuclear magnetic resonance (NMR), and enzyme-linked immunosorbent assay (ELISA). Appropriate methods of detection may depend on the molecular type (e.g., nucleic acid, protein, lipid, etc.) of the biomarker being measured and are readily apparent to one of skill in the art.

[0361] Subjects and Samples

[0362] A subject that is identified, diagnosed, monitored, and / or treated according to the methods described herein may present one or more features of a metabolic syndrome, such as obesity, diabetes (e.g., Type 2 diabetes), insulin resistance, low HDL cholesterol, high LDL cholesterol, or higher than normal blood glucose levels, or subjects with a family history of MASLD and / or liver fibrosis, elevated liver enzyme levels, and / or liver steatosis. Subjects may also have a non-alcoholic steatohepatitis activity (NAS) score >4 (e.g., an at-risk MASH subject).

[0363] A subject that is diagnosed according to the diagnostic methods described herein can also be one that is enrolled in a clinical trial for a new therapy for MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) and the methods can be used to determine an effect on the new therapy in the subject. A subject that is monitored according to the methods described herein may be undergoing or have previously undergone treatment for MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). For example, a subject receiving a MASLD therapy or a liver fibrosis therapy may be monitored for treatment efficacy or disease progression. Alternatively, a subject monitored according to the methods described herein may have been previously diagnosed as having, or is considered at risk of developing, MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). A subject is considered at risk if they present one or more features of a metabolic syndrome, such as obesity, diabetes (e.g., Type 2 diabetes), insulin resistance, low HDL cholesterol, high LDL cholesterol, a higher than normal blood glucose level, a family history of MASLD and / or liver fibrosis, elevated liver enzyme levels, and / or liver steatosis. A subject may have a NAS score >4 (e.g., the subject is an at-risk MASH subject).

[0364] A sample (e.g., a biological sample, e.g., whole blood, plasma, or serum) containing a biomarker described herein (e.g., a biomarker listed in Tables 1 A-1D) can be obtained using methods well known in the art. For instance, samples from a subject may be obtained by venipuncture or from blood, such as serum or plasma. The ease of collection of such samples makes this diagnostic approach available to many more patients than previous diagnostic methods, such as liver biopsy, as diagnosis can be performed non-invasively using blood samples (e.g., serum or plasma samples), such as samples collected during an annual physical exam.

[0365] Methods of Stratifying a Subject for Treatment

[0366] The present disclosure features methods for stratifying a subject (e.g., a subject having MASLD, e.g., MAFL, MASH, at-risk MASH, or cirrhosis) for treatment. Methods of stratifying a subject for treatment may include determining a change (e.g., an increase or decrease >5%) in a level of at least one biomarker (e.g., one, two, three, four, five, six, seven, eight, nine, ten, or more biomarkers) in a biological sample obtained from the subject relative to a reference level.

[0367] Methods of stratifying a subject for treatment may further include classifying the status of the subject’s MASLD as MAFL, MASH, at-risk MASH, or cirrhosis based on the determined change (e.g., an increase or decrease >5%) in the level of the at least one biomarker. For example, an increased level of at least one biomarker relative to the reference level classifies the status of the subject’s MASLD as at- risk MASH, wherein the at least one biomarker selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3- indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158. In some embodiments, the at least one biomarker is selected from 3-ureidopropionate, kynurenine, and a-ketoglutarate.

[0368] Additionally, or alternatively, a decreased level of at least one biomarker relative to the reference level classifies the status of the subject’s MASLD as at-risk MASH, wherein the at least one biomarker selected from the group consisting of: 2'-deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.

[0369] Methods of stratifying a subject for treatment may further include classifying the subject as a candidate to receive a particular therapy (e.g., a MASLD therapy or a liver fibrosis therapy), or a noncandidate for a particular therapy (e.g., a MASLD therapy or a liver fibrosis therapy), based on the classified status of the subject’s MASLD (e.g., based on the status being MAFL, MASH, at-risk MASH, or cirrhosis). For example, a subject’s MASLD classified at-risk MASH classifies the subject as a candidate to receive at least one MASLD therapy selected from the group consisting of: an anti-inflammatory agent, an anti-apoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a peroxisome proliferator-activated receptor (PPAR) modulator, a thyroid receptor beta agonist, an incretin receptor agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a farnesoid X receptor (FXR) agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), JKB-121 , and / or resmetirom (REZDIFFRA™). If the subject’s MASLD is not classified as at-risk MASH, the subject may be classified as a non-candidate for resmetirom (REZDIFFRA™) and other drugs not approved for used during F1 or F4 fibrosis.

[0370] Computer Implemented Methods, Programs, and Devices

[0371] The diagnostic methods and treatments described herein can also be guided by computer implemented models and devices. A machine learning algorithm can be a useful tool to analyze patient data, develop robust, highly-precise models to accurately diagnose a subject as having or not having a MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) and / or to accurately stage the MASLD disease state (e.g., as MAFL, MASH, at-risk MASH, MASH F2, MASH F3, MASH F4, MASH F2-F3, MASH F3-F4, or liver fibrosis).

[0372] A computer implemented method for detection of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject may include training a predictive computer model based on one or more biomarkers (e.g., one or more biomarkers described herein, e.g., see Tables 1A-1 D) or using a model that has already been trained based on the one or more biomarkers in a reference population (e.g., subjects having a known MASLD disease state and / or stage). In some embodiments, the biomarkers are known to be associated with MASLD. In some embodiments, the biomarkers are not known to be associated with MASLD. A predictive computer model may have a pre-set selection of biomarkers to detect a MASLD, and / or a variable selection of biomarkers. The biomarkers used in training a predictive computer model may include metabolites, lipids, glycans, fatty acids, and / or hormones described herein (e.g., see Tables 1A-1 D). In some embodiments, a predictive computer model may incorporate clinical variables in addition to biomarker information.

[0373] A computer implemented method to detect a MASLD disease state may also be used to diagnose a subject as having or not having MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). Information regarding a subject, including values of biomarkers and / or clinical data obtained from a subject may be received by a computing device. In some embodiments, information regarding a subject may be received by the computing device through a graphical user interface. For example, a medical practitioner may input patient information into a computing device. In some embodiments, information regarding a subject may be automatically received by the computing device. In some embodiments, the computing device includes a predictive computer model. In some embodiments, the computing device is in network connection with a second computing device which includes a predictive computer model. When the computing device receives information regarding a subject, the predictive computer model may determine a risk probability score. A risk probability score indicates the likelihood of a subject having or not MASLD (e.g., MAFL or MASH, e.g., at- risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). In some embodiments, the risk probability score is further determined to be a binary score. For example, the predictive computer model may determine if a subject has or does not have MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). In some embodiments, the risk probability score is communicated by the computing device (e.g., on a graphical user interface). A computer implemented diagnostic device may be used to determine the risk of MASLD disease, or disease progression (e.g., MAFL to MASH, MASH to fibrosis, or MASH to at-risk MASH).

[0374] The diagnostic methods and treatments described herein can also be guided by a computer program product (e.g., software). For example, a computer program product (e.g., software) may be used for the detection of MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject. Additionally, it may include a training program based on one or more biomarkers (e.g., one or more biomarkers described herein, e.g., see Tables 1A-1D) or a program that has already been trained based on the one or more biomarkers in a reference population (e.g., subjects having a known MASLD disease state and / or stage). In some embodiments, the biomarkers are known to be associated with MASLD. In some embodiments, the biomarkers are not known to be associated with MASLD. A computer program product (e.g., software) may have a pre-set selection of biomarkers and / or a variable selection of biomarkers described herein. The biomarkers used in the computer program product (e.g., software) may include a combination of the metabolites, lipids, glycans, fatty acids, and / or hormones described herein (e.g., see Tables 1A-1 D). In some embodiments, a predictive computer program product may incorporate clinical variables in addition to biomarker information.

[0375] The computer program product (e.g., software) to detect a MASLD disease state may also be used to diagnose a subject as having or not having MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). Information regarding a subject, including values of biomarkers described herein and / or clinical data obtained from a subject may be received by the computer program product (e.g., software). In some embodiments, information regarding a subject may be received by the computer program product (e.g., software) through a graphical user interface. For example, a medical practitioner may input patient information into the computer program product (e.g., software). In some embodiments, information regarding a subject may be automatically received by the computer program product (e.g., software). When the computer program product (e.g., software) receives information regarding a subject, the predictive computer model may determine a risk probability score. A risk probability score indicates the likelihood of a subject having or not MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). In some embodiments, the risk probability score is further determined to be a binary score. For example, the predictive computer program product (e.g., software) may determine if a subject has or does not have MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). In some embodiments, the risk probability score is communicated by the computer program product (e.g., software). A computer program product may be used to determine the risk of MASLD disease, or disease progression (e.g., MAFL to MASH, MASH to fibrosis, or MASH to at-risk MASH).

[0376] The diagnostic methods and treatments described herein can also be guided by a computer processor (e.g., a computer processor on a non-transitory medium) that can be used to run a program (e.g., computer program product, e.g., software) for detecting or determining MASLD disease status and / or to store data from a sample from a subject. The processor (e.g., computer processor on a non- transitory medium) for use in detecting or determining MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4) in a subject may run a trained computer program (e.g., computer program product, e.g., software) based on the one or more biomarkers (e.g., one or more biomarkers described herein, e.g., see Tables 1A-1 D) in a reference population (e.g., subjects having a known MASLD disease state and / or stage). In some embodiments, the biomarkers are known to be associated with MASLD. In some embodiments, the biomarkers are not known to be associated with MASLD. The processor may store data from a sample from a subject wherein, optionally, the data can be used to further train the computer program. The biomarkers used in training a computer program product may include metabolites, lipids, glycans, fatty acids, and / or hormones described herein (e.g., see Tables 1A-1D).

[0377] The diagnostic methods and treatments described herein can also be guided by a device (e.g., a point of care (POC) device) that can be used to run a program (e.g., computer program product, e.g., software) for detecting or determining MASLD disease status. For example, the device (e.g., POC device) may be used to detect a MASLD disease state and / or lack of disease presence in a subject as having or not having MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). Information regarding a subject, including values of biomarkers (e.g., one or more biomarkers described herein, e.g., see Tables 1 A-1D) and / or clinical data obtained from a subject may be received by the device (e.g., POC device). In some embodiments, information regarding a subject may be received by the device through a graphical user interface. For example, a medical practitioner may input patient information into the device (e.g., POC device). In some embodiments, information regarding a subject may be automatically received by the device (e.g., POC device). When the device receives information regarding a subject, a predictive computer model may determine a risk probability score. A risk probability score indicates the likelihood of a subject having or not MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). In some embodiments, the risk probability score is further determined to be a binary score. Thus, the device (e.g., POC device) may determine if a subject has or does not have MASLD (e.g., MAFL or MASH, e.g., at-risk MASH) and / or liver fibrosis (e.g., MASH F2-F3, MASH F3-F4, or MASH F4). In some embodiments, the risk probability score is communicated by the device to the user.

[0378] Machine Learning Coding and Commands

[0379] The following commands and coding are non-limiting and may be used with the machine learning models described herein. Examples in how machine learning can be programmed to deliver the results described herein are provided by the codes below and could also be provided by similar ones, which a person knowledgeable in the art and / or science of machine learning and MASH could provide to a computer to process the provided data and, thus, prompt delivery of information for determining whether a liver condition (such as liver disease, e.g., MASH F2-F3 stage or F4 stage) is present or not. One of skill in the art would understand that modifications may be made to the following codes and commands to change the inputs and outputs, where and as needed.

[0380] Modeltest library(catboost) library(dplyr) library (pROC) column_of_interest <- "NASHF2F3" top_features <- "alpha-ketoglutarate" train ing_select <- training %>% select(all_of(c(top_features, column_of_interest))) validation_select <- validation %>% select(all_of(c(top_features, column_of_interest))) full_data <- bind_rows(training_select, validation_select) train_pool <- catboost.load_pool( data = training_select %>% select(-all_of(column_of_interest)), label = training_select[[column_of_interest]] valid_pool <- catboost.load_pool( data = validation_select %>% select(-all_of(column_of_interest)), label = validation_select[[column_of_interest]] full_pool <- catboost.load_pool( data = full_data %>% select(-all_of(column_of_interest)), label = full_data[[column_of_interest]] param_grid <- list( iterations = c(200, 400, 600), learning_rate = c(0.05, 0.075, 0.1 , 0.2), depth = 5:9,

[0381] I2_leaf_reg = 1 :6 best_params <- grid_search_catboost(train_pool, valid_pool, param_grid) model <- catboost.train( params = best_params, learn_pool = train_pool, test_pool = valid_pool results <- evaluate_model(model, full_data, top_features, column_of_interest) pred_prob <- results$predicted_probabilities[, 2] threshold <- results$optimal_threshold pred_class <- ifelse(pred_prob >= threshold, 1 , 0) actual <- results$roc_curve$response

[0382] TP <- sum(pred_class == 1 & actual == 1 )

[0383] TN <- sum(pred_class == 0 & actual == 0)

[0384] FP <- sum(pred_class == 1 & actual == 0)

[0385] FN <- sum(pred_class == 0 & actual == 1 ) metrics <- c(

[0386] AUC = results$auc, Accuracy / length(actual), Sensitivity FN),

[0387] Specificity FP), PPV = FP),

[0388] NPV FN) base_name <- gsub(" A-Za-zO-9]+", paste(c(column_of_interest, top_features), collapse = " ")) catboost.save_model(model, pasteO(base_name, ".cbm")) write. csv(best_params, pasteO(base_name, "_params.csv"), row.names = FALSE) print(round(metrics, 3))

[0389] Feature Selection library(dplyr) library(catboost) library(caret) library (pROC) library(glmnet)

[0390] Iibrary(ggplot2) library(ggrepel) column_of_interest <- "NASHF2F3" catboost_params <- list(loss_function = "Logloss", iterations = 200, learning_rate = 0.05, depth = 6, I2_leaf_reg = 4, custom_metric = "AUC, eval_metric = "AUC", random_seed = 2, od_type = "Iter", bootstrap_type = "MVS", od_wait = 20, verbose = 100) predetermined_vars <- cf'ALTUIL", "InsulinemiamlUml", "ASTUIL", "BMI") testco <- read.csv("testco2 - Copy.csv", check.names = FALSE) train_df <- filter(testco, Set == "Train") validation_df <- filter(testco, Set == "Validation") write. csv(train_df, "training_clinical_model.csv", row.names = FALSE) write. csv(validation_df, "validation_clinical_model.csv", row.names = FALSE) training <- select(train_df, {{ column_of_interest }}, everythingQ) validation <- select(validation_df, {{ column_of_interest }}, everything()) candidate_cols <- c("Age", "sexI fOm", "BMI", "waist_hip", "DiabetesYESI NOO", "InsulinemiamlUml", "Plateletsxl 091", "ASTUIL”, "ALTUIL", "GAMMAGTUIL", "Albumingl", "TotalMetS", "GlucoseMetS", "HDLMetS", "TGMetS", "HypertensionMetS", "WaistMetS", "Follistatinngml", "FSTL3ngml", "ActivinBpgml", "ActivinApgml", "FreelGFIngml", "TotalGDF15pgml", "lntactGDF15pgml", "Adiponectinpgml", "Leptinngml") training <- select(training, {{ column_of_interest }}, all_of(candidate_cols)) validation <- select(validation, {{ column_of_interest }}, all_of(candidate_cols)) predictors <- select(training, -all_of(column_of_interest)) %>% select_if(is. numeric) outcome <- training[[column_ofjnterest]] summaryFunction <- function(data, lev = NULL, model = NULL) postResample(data$pred, data$obs) fitFunction <- function(x, y, first, last, ...) { x <- as.matrix(data.frame(lapply(x, as. numeric))) pool <- catboost.load_pool(x, label = y) catboost.train(learn_pool = pool, params = list(loss_function = "Logloss", iterations = 100))

[0391] } predFunction <- function(model, x) { pool <- catboost.load_pool(as.matrix(data.frame(lapply(x, as. numeric)))) catboost.predict(model, pool)

[0392] } rankFunction <- function(model, x, y) { imp <- catboost.get_feature_importance(model) data.frame(var = colnames(x), importance = imp)

[0393] } selectSizeFunction <- function(x, metric, maximize) min(x[which.max(x[, metric]), "Variables"]) selectVarFunction <- function(y, size) head(y[order(-y$importance), "var"], size) catboostFuncs <- list(summary = summaryFunction, fit = fitFunction, pred = predFunction, rank = rankFunction, selectSize = selectSizeFunction, selectVar = selectVarFunction) set.seed(1231 ) control <- rfeControl(functions = catboostFuncs, method = "cv", number = 10) results <- rfe(x = predictors, y = outcome, sizes = 1 :10, rfeControl = control) imp_rfe <- aggregate(importance ~ var, results$variables, mean) max_mag <- function(loadings) apply(loadings[, 1 :2], 1 , function(v) sqrt(sum(vA2))) prep_pca <- function(df, feats) { scaled <- scale(select(df, all_of(feats))) prcomp(scaled, center = TRUE, scale. = TRUE)

[0394] } pca_res <- prep_pca(training, colnames(predictors)) imp_pca <- data.frame(Feature = rownames(pca_res$rotation), Magnitude = max_mag(pca_res$rotation))

[0395] Mode <- function(x) unique(x)[which.max(tabulate(match(x, unique(x))))] imp_na <- function(df) df %>% mutate_if(is. numeric, ~ifelse(is.na(.), median)., na.rm = TRUE), .)) trainjmp <- imp_na(training) x_train <- as.matrix(select(train_imp, -all_of(column_of_interest))) y_train <- train_imp[[column_of_interest]] alpha_grid <- seq(0.1 , 0.9, 0.1 ) cvfits <- lapply(alpha_grid, function(a) cv.glmnet(x_train, y_train, alpha = a, family = "binomial")) alpha_best <- alpha_grid[which.min(sapply(cvfits, function(cv) min(cv$cvm)))] lambda_best<- cvfits[[which.min(sapply(cvfits, function(cv) min(cv$cvm)))]][["lambda.min"]] coef_df <- as.data.frame(as.matrix(coef(glmnet(x_train, y train, alpha = alpha_best, lambda = lambda_best, family = "binomial")))) coef_df <- tibble::rownames_to_column(coef_df, "Feature") %>% filter(Feature != "(Intercept)") %>% rename(Coefficient = 's0') %>% mutate(AbsCoefficient = abs(Coefficient)) norm <- function(v) v / max(v, na.rm = TRUE) imp_rfe <- mutate(imp_rfe, norm_rfe = norm(importance)) imp_pca <- mutate(imp_pca, norm_pca = norm(Magnitude)) coef_df <- mutate(coef_df, norm_en = norm(AbsCoefficient)) combined <- fulljoin(imp_rfe, imp_pca, by = c("var" = "Feature")) %>% f ulljoin(coef_df , by = c("var" = "Feature")) %>% mutate(score = rowSums(select(., starts_with("norm_")), na.rm = TRUE)) %>% arrange(desc(score)) n keep <- 20 selected <- head(combined, n keep) write. csv(selected, "top_selected_features_per_combined_score.csv", row.names = FALSE) features_final <- union(predetermined_vars, selected$var) training_final <- select(training, all_of(c(features_final, column_of_interest))) validation final <- select(validation, all_of(c(features_final, column ofjnterest))) train_pool <- catboost.load_pool(select(training_final, -all_of(column_of_interest)), label = training_final[[column_of_interest]]) valid_pool <- catboost.load_pool(select(validation_final, -all_of(column_of_interest)), label = validation_final[[column_ofjnterest]]) model_final <- catboost.train(params = catboost_params, learn_pool = train_pool , test_pool = valid_pool) save_name <- gsub("[AA-Za-zO-9]+", paste(c(column_of_interest, "FS"), collapse = " ")) catboost.save_model(model_final, pasteO(save_name, ".cbm")) write. csv(catboost_params, pasteO(save_name, "_params.csv"), row.names = FALSE)

[0396] EXAMPLES

[0397] The following examples are put forth so as to provide those of ordinary skill in the art with a description of how the compositions and methods described herein may be used, made, and evaluated. These examples are intended to be purely exemplary of aspects of the disclosure and are not intended to limit the scope of what the inventors regard as their invention.

[0398] Example 1. A multinational study of serum metabolomics and lipidomics in the pathophysiology and non-invasive diagnosis of metabolic dysfunction-associated steatotic liver disease through accurate, lightweight, machine learning models

[0399] In a previous study (e.g., see U.S. International Patent Application No. WO 2021 / 092265, which is incorporated herein by reference in its entirety), we identified panels of biomarkers that could diagnose metabolic dysfunction-associated steatotic liver disease (MASLD), metabolic dysfunction-associated steatohepatitis (MASH), and fibrosis. We sought to streamline and further amplify our previous study by using state-of-the-art machine learning algorithms and automated software pipelines to create new, robust, highly-precise models with the simplest possible structure and the highest possible score. The present example describes our multinational metabolomic and lipidomic study aimed at identifying additional biomarkers of MASLD, as well as building, training, and validating a robust, lightweight, and non-invasive model for the detection of MASLD, those at risk of developing of MASLD, or those at risk of MASLD disease progression (MAFL, MASH, at-risk MASH, or liver fibrosis). Additional aims of this study include:

[0400] • mapping the clinical and serum biochemical, hormonal, and metabolipidomic signatures across the entire MASLD histological spectrum, with emphasis on (i) the NAS composite score and the fibrosis score, and (ii) binary histological outcomes (e.g., any MASLD, any MASH, any fibrosis, stage 2 fibrosis (F2), stage 3 fibrosis (F3), stage 4 fibrosis (F4), and at-risk MASH);

[0401] • identifying any differentially regulated or co-regulated metabolic and lipid pathways across binary outcomes of MASLD in conjunction with clinical and hormonal data;

[0402] • implementing the models building and validation approaches across additional binary outcomes of MASLD;

[0403] • benchmarking model performance across clinically and demographically relevant subgroups within the study population; and

[0404] • comparing the models performance with 15 applicable non-invasive indices of MASLD stratification.

[0405] The study population (n=443) of the present Example was pooled across the following three cohorts. Detailed inclusion and exclusion criteria by study protocol are presented in Table 2.

[0406] • 296 inpatients from a Bariatric-Metabolic Surgery Clinic of the Universita Cattolica del Sacro Cuore in Rome, Italy; • 30 outpatients from a Gastroenterology-Hepatology Clinic of the University of Athens Laikon Hospital, in Athens Greece; and

[0407] • 117 outpatients from a Gastroenterology-Hepatology Clinic of the University of Sydney, Westmead Hospital, Sydney.

[0408] Table 2. Inclusion and Exclusion Criteria Across Study Populations

[0409] As described in more detail below, we created two main categories of models: (1 ) conservative models, which include predetermined domain knowledge variables pre-configured and the first additional top-performing variable from our feature selection process; and (2) main models, which implement floating feature selection, a sophisticated technique that further expands variable selection within a predetermined range of top-performing variables (e.g., top 5 or top 10), to sequentially assess aggregate models and find the best possible additional variables that maximize model performance. Model structures and characteristics are shown in Table 3 below and report their cross-validated and resampled aggregate scores and Youden’s index paired sensitivity-specificities, to ensure robustness and generalizability notwithstanding data distribution, and including an additional layer of leave-one-out cross- validation to gauge model performance at the individual level considering the different distributions of histological characteristics in our cohort. Robust metabolites and lipids used in the model are outlined in bold. Table 3. Model Component and Aggregate Cross-validated and Resampled Performance Metrics Across the Validation Set and Entire Cohort

[0410] ALT=Alanine aminotransferase; AST=Aspartate aminotransferase; AUROC=Area under the receiver operating characteristics curve; BMI=body mass index; Cl=confidence intervals; FA=fatty acid; NPV=negative predictive value; PC=phosphatidylcholine; PPV=positive predictive value; TAG=triacylglycerol; Total MetS=no. of total components of the metabolic syndrome. Results

[0411] Pooled clinical, biochemical, hormonal, metabolomic, and lipidomic data across three countries

[0412] Percutaneous ultrasonography-guided hepatic biopsies were performed after the exclusion of other hepatic diseases or causes of fatty hepatic infiltration. Two independent expert pathologists assessed the biopsies evaluating the non-alcoholic activity score (NAS) according to NASH Clinical Research Network (NASH CRN) criteria, with a degree of concordance of 85%. MASH was defined as NAS >4, including a score of >1 in each NAS component (e.g., steatosis, hepatocellular ballooning, and lobular inflammation). Hepatic fibrosis was evaluated with stage 0 (F0), which indicates no fibrosis; stage 1 (F1), which indicates centrilobular pericellular fibrosis (F1); stage 2 (F2), which indicates centrilobular and periportal fibrosis; stage 3 (F3), which indicates bridging fibrosis; and stage 4 (F4) which indicates cirrhosis. At-risk MASH was set as the primary histopathological outcome and defined as NAS>4 with all components >1 , and fibrosis >2.

[0413] Demographic and clinical data by study center are presented in Table 4. For pathophysiology- related analyses, the entire cohort was analyzed. For the creation and validation of models, the cohort was split into a training set (80%, n=353) and a hold-out, independent validation set (20%, n=90). Cohort separation was conducted at random but while ensuring a uniform distribution of at-risk MASH and type of clinic (e.g., bariatric vs gastroenterology / hepatology), between the sets, which also resulted in uniform clinical, biochemical, and overall histological characteristics.

[0414] Table 4. Demographic, Clinical, and Biochemical Data of Participants in the Multi-center, Liver

[0415] Biopsy-based Study.

[0416] Alanine transaminase (ALT); Aspartate aminotransferase (AST); Body mass index (BMI); Total chole Gamma-glutamyl transferase (GGT); Hemoglobin A1c (HbA1c); High-density lipoprotein cholesterol density lipoprotein cholesterol (LDL-C); Metabolic dysfunction-associated steatotic liver disease (MASL Metabolic dysfunction-associated steatohepatitis (MASH); Metabolic syndrome (MetS); Non-alcoholic steatohepatitis activity score (NAS). P-values for continuous variables were obtained through Students' t-test for independent samples. P-values for categorical variables were obtained through Chi-squared test.

[0417] Overall study flow, including the training-validation split and applicable measurements are shown within FIG 1A, FIG. 1B, and FIG. 1C. We acquired serum metabolomic and lipidomic profiles from all participants using the latest platforms from Metabolon Inc. using ultra-high performance liquid chromatography / tandem mass spectrometry, and incorporating 839 metabolites and 840 lipids in the analysis following data filtering. These measurements were supplemented with routine clinical and biochemical indices and ten hormones (e.g., see Table 5). Table 5. Hormones and Kit Information

[0418] We calculated 15 predefined non-invasive indices of MASLD stratification which were optimized for our data and compared, using both DeLong’s tests and repeated / resampled cross-validation, with our machine learning models across all binary outcomes. The overall statistical workflow and machinelearning pipeline is shown in FIG. 1D and FIG. 1E. A list of metabolite and lipids and pathway designations were identified. Several detected unknown metabolites relevant to the analysis shown in Table 6. Table 6. Available Data an Unknown Metabolites Markedly Regulated

[0419] To capture the dimensionality of the overall circulating metabolipidomic signatures, we performed principal component analyses (PCA) of the top 100 metabolites and lipids by variance across either NAS or fibrosis score, following the Benjamini-Hochberg false discovery rate (BH-FDR) correction for multiple comparisons at a threshold of p<0.05 (FIG. 1F and FIG. 1G). Metabolomic profiles were distinct between country-specific populations, demonstrating a clear differentiation across NAS score progression as well as the primary outcome of at-risk MASH, with the prime drivers of this being, for example, mannonate and glycuronate (both microbial metabolites) in the direction of bariatric patients from Italy, as well as succinate, aspartate, choline, proglutamylglycine (5-L-Glutamylglycine) and 3-ureidopropionate (3-UPA) (chiefly amino acids and nucleotides) in the general direction of gastrointestinal patients and at-risk MASH (FIG. 1 F). On the contrary, lipid profiles were uniform across cohorts and histological outcomes, whereas the examination of top variances revealed a triglyceride surge chiefly governed by 48-52 carbon triglycerides containing 16-18 carbon fatty acids of variable saturation (FIG. 1 G). By extrapolating all significant metabolomic and lipidomic variables used in the PCAs, we ran a partial least-squares discriminant analysis across the entire cohort to assess the contribution of features in the identification of at-risk MASH (FIG. 1H), revealing a lack of lipids as the primary separating variables in component 1 and underscoring the significance of N-acetylmethionine, sphingolipids, mannonate, amino-acids and 3-UPA as the principal features per variable importance projection scores.

[0420] Distinct metabolipidomic signatures define MASLD progression differentially across fibrosis and MASH The overall variances of all metabolites and lipids across the fibrosis score and the NAS composite score are shown in FIG. 11 and FIG. 1 J. For metabolomics, there was a bidirectional surge of mannonate, sphingolipids (e.g., sphingosine, sphionganine, sphingadienine), branched-chain amino acids (BCAAs) and nucleotides (e.g., xanthine, 2-deoxyuridine, and 3-UPA), and a differential regulation of unknown metabolites (e.g. X-23654 and X-12456, chiefly upregulated in fibrosis) and choline, which was primarily regulated under NAS. Lipid profiles displayed a uniform, bidirectional surge of triglyceride species that eclipsed all other lipid variabilities, except for ceramides and their subspecies which were markedly changed in NAS. Triglycerides followed a uniform trend that overshadowed all other lipid species, both due to the sheer number of triglyceride species (515 of the total 840 lipids), and their unique directionality, indicating a consistent upregulation post-fibrosis 1 -2 and a steep upregulation in definite MASH (NAS>4).

[0421] To fully examine the circulating metabolipidomic signatures and plot their progression across the disease in conjunction with known clinical and biochemical traits, we created a heatmap of the top 100 proteins per joint NAS / fibrosis variability, the top 50 lipid species likewise assessed following the omission of triglycerides in view of their uniformity, and all relevant clinical, biochemical, and hormonal variables. We further showcase molecule / metric type and whether variables were significantly and differentially regulated across pre-determined binary histological stages following BH-FDR (e.g., see FIG. 2A, FIG. 2B, FIG. 2C, FIG. 2D, FIG. 2E, FIG. 2F, FIG. 2G, FIG. 2H and FIG. 2I) in conjunction with composite histological data (FIG. 2J). Unsupervised clustering of sub-heatmaps revealed three distinct progression profiles that were uniform for the top metabolites, lipids, and clinical features (FIG. 2K, FIG. 2L and FIG. 2M). The green cluster consisted of molecules that were higher in healthy individuals and principally downregulated at the late-stages of the disease, encompassing chiefly sulfate molecules belonging to steroidogenesis pathways, guanosine, inosine, N-acetylmethionine, fibrinopeptides, phosphatidylcholine (e.g., PC(18:2 / 20:1 )), adiponectin, and HDL. The blue cluster was principally upregulated in the intermediate-to-late stages of the disease, and comprised, amongst others, BCAAs and their metabolites, xenobiotic metabolites (e.g., kynurenine, mannonate), glutamate and purine metabolites, and three unknown molecules. These were mirrored by a constellation of monoacylglycerols and ceramides, accompanied by one diacylglycerol and one lysophosphatidylcholine. Most lipids within this cluster contained saturated fatty acids and, notably, did not fully mirror the motif of metabolites at late histological stages (Fig. 21). The clinical variables more closely mirroring these patterns were chiefly total and LDL cholesterol, anthropometric indices and body mass index (BMI), the metabolic syndrome and its components, insulinemia, and the homeostatic model of insulin resistance (HOMA-IR). Finally, the red cluster contained hepatotoxic molecules that were robustly elevated in the late stages of the disease, notably cirrhosis. Bile acid, pantothenate, TCA cycle metabolites, several nucleotides (e.g., 3-UPA and xanthine), sphingolipid intermediates, and a considerable number of unknown metabolites were upregulated after F>3 and within the histological definition of MASH. These were mirrored by most triglycerides as well as several mono- and polyunsaturated diacylglycerols, a few ceramides, and a small number of phosphatidylethanolamines and phosphatidylcholines. Clinically, this pattern was chiefly reflected by transaminases (aspartate tramsaminase, AST; alanine transaminase, ALT), gamma-glutamyl transfsease, components of the activin-follistatin axis, total and intact GDF-15 (a known proinflammatory mitokine), leptin (an indicator of adiposity), Fibrosis-4-Score (FIB-4) and its components, fasting glucose, diabetes, and glycosylated hemoglobin.

[0422] To further ascertain underlying associations between all variables with histology, we performed and plotted spearman correlations of variables with either NAS or fibrosis scores, following BH-FDR correction (FIG. 2N and FIG. 20). Component-specific associations for NAS are shown within. Metabolites and liver-specific clinical features were more robustly associated with fibrosis, whereas lipids and clinical variables with NAS. It became apparent that several metabolites (e.g., sphingolipids, s- adenosylhomocysteine, glucamate, and 3- UPA) followed the pro-fibrotic, hepatotoxic trajectories delineated by transaminases and GDF-15, whereas lipid species, in particular 16-carbon saturated and monounsaturated diacylglycerols, alongside insulinemia and the metabolic syndrome, were more closely associated with MASH. Notably, the variables most negatively associated with either component were n- acetylmethionine, inosine, guanosine, and several fibrinopeptide components.

[0423] Distinct metabolipidomic signatures define binary histopathological outcomes of MASLD To fully elucidate the underlying signatures governing clinically-relevant MASLD thresholds, we performed unpaired Welch’s t-tests of metabolites across binary histopathological outcomes (subtracting Iog2-transformed values of patients with - patients without the outcome) across both metabolites and lipid species, followed by pathway analyses (FIG. 3). MASLD was characterized by a downregulation of sulfates and fibrinopeptides, a mild increase in sphingolipids and fatty acid metabolites as well as components of the TCA cycle and BCAA metabolites. We also ascertained an upregulation of long-chain triglycerides, a notable reduction in dihydrocylceramides and monoacylglycerols, and absolute ratios of MUFA / PUFA av5 w3 / w6 <1 (FIG. 3A, FIG. 3B, FIG. 30, and FIG. 3D). Most of these changes persisted in MASH wherein we also observed a further upregulation of sphingosines, taurocholate and bile acid metabolism intermediates, whereas the triglyceride surge became apparent, notably accompanied by a downregulation of phosphatidylcholines and contrary with MASLD an upregulation of ceramides and their surrogates with the exception of DCER(26:1 ), as well as an increase of diacylglycerols and cholesterol esters, and a further increase in saturated fatty acids (FIG. 3E, FIG. 3F, FIG. 3G, and FIG. 3H). The histological definition of fibrosis entailed an even steeper upregulation of sphingosines and overfall similar metabolite changes as observed in MASLD, notably without entailing bile acid metabolism, and further enriched by glutamate and polyuamine metabolites. It also revealed a lipidomic profile similar to MASH, with upregulation of triglycerides and diacylglycerols, ceramides (particularly dihydro- and hexocylceramides) (FIG. 31, FIG. 3J, FIG. 3K, and FIG. 3L). Significant fibrosis similarly signified a surge in bile acid metabolism and lipid synthesis, as well as purine, glutamate, and tryptophan metabolites. Lipidomics revealed a similar increase in tri- and diacylglycerols, and an upregulation of dihydroceramides with concurrent downregulation of lactocylceramides and phosphatidylcholines, and an upregulation, for the first time, of 18-carbon fatty acids (FIG. 3M, FIG. 3N, FIG. 30, and FIG. 3P). Notably all disease states entailed the upregulation of palmitic (16:0) and palmitoleic acid (16:1 ), stearic (18:0) and oleic acid (18:1 ), and various eicosanoids including arachidonic acid (20:4). At-risk MASH for the first time demonstrated differential profiles, primarily focused on the upregulation of sphingosines, s- adenosylhomocysteine, glutamate and 3-LJPA, bile acid, BCAA and TCA metabolites but also entailing sulfates implicated in the production of androgenic steroids, sphingolipids, pentose and phenylalanine metabolites, and importantly purine and pyrimidine metabolism, the metabolites of which pathways were both up- and down-regulated. In lipidomics, while the triglyceride surge persisted, there was a notable and steep downregulation of ceramides (dihydroceramides in particular) and monoacylglycerols, and a lack of w3 / w6 enrichment. (FIG. 3Q, FIG. 3R, FIG. 3S, and FIG. 3T). Lastly, advanced fibrosis revealed a metabolic substrate governed by several non-characterized, unknown molecules, and partially mirroring the changes observed in at-risk MASH, whereas lipidomic changes were blunted because of the observed plateau in triacylglycerol levels during late stages of the disease, indicating a downregulation of mono- and diacylglycerols as well as ceramides that were upregulated in the initial stages of MASLD (FIG. 3U, FIG. 3V, FIG. 3W, and FIG. 3X).

[0424] Metabolipidomic pathway co-regulation in at-risk MASH and across MASLD histology

[0425] To integrate the observed metabolipidomic changes in the context of co-regulation and conventional clinical and biochemical indices, we performed intra-omics spearman correlations and weighted network correlation analyses of omics with clinical variables for each of the binary states (e.g., any MASLD, any MASH, any fibrosis, stage 2 fibrosis (F2), stage 3 fibrosis (F3), stage 4 fibrosis (F4), and at-risk MASH) (FIG. 4, FIG. 9, FIG. 10, FIG. 11 , and FIG. 12). As an example, the regulation of metabolites under at-risk MASH was apparent in five distinct clusters (FIG. 4A). Cluster 1 represented mildly upregulated molecules throughout the entire spectrum of at-risk MASH, chiefly pertaining to tryptophan and BCAA metabolism. Cluster 2 contained bile acid and fatty acid metabolites, whereas Cluster 3 was principally upregulated in the late stages of the disease and contained phingosines, long- chain and phospholipid metabolites. Cluster 4 was negatively associated with Cluster 3 with its molecules following an opposite direction, chiefly pertaining to pentose and early pyrimidine metabolism. Cluster 5 contained fibrinopeptides and other molecules with intermediate relative abundances. Lipid regulation was principally governed by triglycerides (FIG. 4B). Interestingly, triglycerides with under 50 total carbons clustered together, as did triglycerides with over 50 total carbons alongside 16-carbon diacylglycerols, further delineating a smaller cluster of 56-carbon triglycerides and a few phosphatidylcholines and phosphatidylethanolamines. To integrate these insights in the context of clinical knowledge, we performed weighted network analyses (WCNA) of significant metabolites and lipids. In the WCNA of metabolites, it became apparent that a cluster of molecules associated with liver damage (e.g., transaminases, follistatin-like 3, and GDF-15) were associated with all metabolite modules, particularly with those containing fatty and bile acid metabolism (FIG. 4C and FIG. 4D). Variables pinpointing impaired glucose homeostasis were associated to a lesser degree with these modules, whereas anthropometries, IGF-1 , and total and LDL cholesterol were negatively associated with bile acid and polyamine metabolites. Lipidomics network analyses were markedly influenced by triglycerides, with clinical associations mirroring this trajectory. The metabolic syndrome and conventionally assessed triglycerides were more robustly associated with lipid modules, followed by indices of glucose homeostasis, and negatively associated with HDL cholesterol, activin B, and adiponectin. Most of the conjugated fatty acids were long- chain, saturated fatty acids, whereas lipid variability was accentuated particularly in the first module, which demonstrated no associations with weight, hip circumference, and fasting insulin (FIG. 4E and FIG. 4F). A similar approach was followed for all other binary thresholds (FIG. 9, FIG. 10, FIG. 11 , and FIG. 12) and notably failed to produce meaningful clusters in the F>3 state.

[0426] Creation and validation of lightweight categorical gradient boosting machine learning models using clinical and metabolipidomic variables We sought to actualize metabolomic and lipidomic insights via a domain-knowledge, physiology- driven approach to build robust and lightweight predictive models for the non-invasive prediction of at-risk MASH and other MASLD outcomes. We implemented a comprehensive, multi-step cross-validated feature selection process and nested cross-validated hyperparameter tuning building models within the training cohort on the basis of pre-defined, domain-knowledge variables enriched with top-performing metabolites / lipids (conservative models, green curves). We further expanded metabolite / lipid selection through a floating sequential feature selection algorithm that added variables to the models from a limited pool of top-performing variables on the basis of performance increase (floating models, blue curves). We evaluated models through k-fold cross-validation, resampled 10 times. K=5 for the independent hold-out validation cohort and 10 for the training and entire cohorts. We holistically examined model performance throughout random data slices in all applicable datasets (FIG. 5 and FIG. 6). Results from the training cohort show that nearly all cross-validated metrics were >90%.

[0427] For at-risk MASH, a model with ALT, AST, and the number of metabolic syndrome components (TotalMetS) as predetermined, domain-knowledge variables, selected 3-UPA as the top-performing metabolite, which exhibited robust performance, achieving a mean cross-validated AUROC of 0.89 (0.87- 0.90), a mean accuracy of 0.87, a mean youden’s sensitivity of 0.93, and a mean youden’s specificity of 0.81 in the cross-validated and resampled validation cohort. The same model also exhibited robust performance, achieving a mean cross-validated AUROC of 0.93, a mean accuracy of 0.89, a mean youden’s sensitivity of 0.89, and a mean youden’s specificity of 0.88 in the entire cohort (FIG. 5A and FIG. 5B). The floating model was further enriched by kynurenine, a tryptophan metabolite (Fig. 5C), achieving a mean cross-validated AUROCs of 0.92 and 0.95 in the validation and entire cohorts, respectively, with mean accuracies of 0.90 and 0.91 , sensitivities of 0.92 and 0.91 , and specificities of 0.87 and 0.89 (FIG. 5D and 5E). Leave-one-out cross-validation also revealed robust metrics (AUROC=0.93, accuracy=0.86, Youden’s sensitivity=0.82, Youden’s specificity=0.89, positive predictive value (PPV)=0.81 , negative predictive value (NPV)=0.9) (FIG. 5F).

[0428] For advanced fibrosis (e.g., F>3), starting from the same domain-knowledge variables, 3-UPA was once again selected, indicating its accentuated role as a threshold of advanced MASLD. The conservative model exhibited robust performance with (i) a mean cross-validated AUROC=0.94 and 0.92 in the validation and entire cohorts, respectively; and (ii) an accuracy, youden’s sensitivity, and youden’s specificity all being >90% in the validation cohort (FIG. 5G and FIG. 5H). Sequential floating selection initially did not pick any new metabolites (FIG. 5I), though alpha-ketoglutarate, which had an equal contribution with 3-UPA, was added on the basis of performance across overall folds. The model remained robust in both the validation and entire cohort, achieving robust AUROCs (0.93) and accuracy (0.94) as well as near-perfect sensitivity (0.99) and very good specificity (0.89) in the validation cohort. The model also achoeved a mean cross-validated AUC of 0.96 and an accuracy of 0.94, Youden’s sensitivities and specificities of 0.93 in the entire cohort, demonstrating exceptional metrics in the leave- one-out cross-validation (FIG. 5K, FIG. 5K, and FIG. 5L). The model for F>2 used diabetes, TotalMetS, insulinemia, and transaminases as domain knowledge variables, selecting the bile acid intermediate 7- HOCA in the conservative approach and attaining good metrics in the cross-validation scheme. The floating algorithm selected kyunurenine, 3-indoleglyoxylic acid, and an unknown metabolite (X-21286), without achieving any further increases in validation performance (FIG. 5M, FIG. 5N, FIG. 50, FIG. 5P, FIG. 5Q, and FIG. 5R).

[0429] For simple fibrosis detection, BMI alongside transaminases and TotalMetS were established as domain knowledge variables, with feature selection further including sphingadienine and training a model with a cross-validated AUROCs of 0.88 and 0.96, accuracies of 0.92 and 0.96, Youden’s sensitivities of 0.88 and 0.91 , and specificities of 0.90 and 0.922, in the validation and entire cohorts, respectively (FIG. 6A and FIG. 6B). Floating selection picked X-12456, a suspected sulfate / steroid, and achieved a mean AUROC of 0.92 and accuracy of 0.93 in the validation cohort, with near-perfect scores in the entire cohort (0.98, 0.98, 0.96, and 0.95 for mean AUROC, accuracy, Youden’s sensitivity, and Youden’s specificity, respectively) and, importantly, the leave-one-out validation (0.98, 0.96, 0.98 and 0.88 for mean AUROC, accuracy, Youden’s sensitivity, and Youden’s specificity, respectively; additionally 0.98 PPV and 0.86 NPV) (FIG. 6D, FIG. 6E, and FIG. 6F). To detect MASH, we pre-selected BMI, fasting insulin, TotalMetS, transaminases, and indolelactate. The floating selection selected phenylalanine and three lipids, including a ceramide (e.g., CER18:0), a phosphatidylcholine (e.g., PC16:0 / 16:1 ) and a triglyceride (e.g., TAG48:0 / FA16:0), achieving a mean cross-validated AUROC of 0.91 in the validation cohort and 0.99 in the entire cohort, as well as mean accuracy of 0.90 and 0.98, respectively, Youden’s sensitivity of 0.88 and 0.98, and Youden’s specificity of 0.92 and 0.99, respectively (FIG. 6G, FIG. 6H, and FIG. 61). Finally, a model capable of detecting MASLD and using BMI, TotalMetS, and transaminases selected mannonate, further enriching its structure with the addition of TAG54:4 / FA22:4 and attaining a mean cross-validated AUROC of 0.93 and 0.98, and accuracy of 0.93 and 0.97 in the validation and entire cohort, respectively (FIG. 6M, FIG. 6N, FIG. 60, FIG. 6P, FIG. 6Q, and FIG. 6R).

[0430] To comprehensively benchmark model performance we also calculated Shapley Additive exPlanations (SHAP) values for each variable, plotted model learning curves at 5-10% slices of the training and validation sets to gauge overtraining, and performed subgroup analyses within the entire cohort without cross-validation, examining model performance in the training, validation, and entire sets, as well as in clinically and demographically relevant subgroups: patients with vs. without diabetes, patients with vs. without obesity, and patients with vs. without MAFLD, wherever applicable, as well as in all patients excluding cirrhosis and patients per clinic type (bariatric vs. gastroenterology / hepatology). For our main at-risk MASH and F>3 models, these benchmarks can be found in FIG. 13 and demonstrate exceptional overall performance. We further calculated fifteen applicable non-invasive indices for MASLD stratification (e.g., FIB4, fatty liver index (FLI), HIS, LAP, NLFS, index of NASH (ION), TYG (two equations), acNASH, fibrotic NASH index (FNI), NAFLD Fibrosis Score (NFS), Aspartate transaminase / Alanine aminotransferase (AST / ALT), ALT / AST, AST to Platelet Ratio Index (APRI), and Homeostatic Model Assessment for Insulin Resistance (HOMA-IR), fully described within Table 7, re-optimized their optimal thresholds across every applicable outcome and cohort, and compared them with our models in the validation and entire cohort, and at a random 90-person slices of the entire cohort, to match validation size.

[0431] Table 7. Performance characteristics of optimized, pre-defined established non-invasive indices in the entire cohort for the detection of At-risk MASH, compared with our CatBoost models

[0432] Pre-calculated index thresholds re-optimized to maximize performance in the entire cohort and to be able to favorably compare with Categorical Gradient Boosting Machines (CatBoost) models. CatBoost conservative, a model with ALT, TotalMetS, AST and 3-ureidopropionate only; CatBoost Main, full model with ALT, TotalMetS, AST, 3-ureidopropionate and kynurenine; CatBoost configured, same as previous but with re-optimized hyperparameters for the validation cohort instead of nested cross-validated grid search within the training cohort.

[0433] The comparisons were effectuated both using 5-fold cross-validation, repeated 10 times, and holistically examined without cross-validation within every applicable group using Delong’s tests. All applicable performance metrics were calculated regardless of outcome distribution to match the characteristics of each cohort, occasionally resulting in overall low PPV or NPV owing to limited distribution of outcome-specific “controls” in smaller cohorts (validation or slices). The benchmarks of our at-risk MASH model can be found in FIG. 14, FIG. 15, and FIG 16 for the entire, validation, and random- sliced sets, respectively. These results demonstrate that our models demonstrate exceptional performance stability and overwhelmingly outperform all applicable calculated indices, which is particularly apparent in cross-validation, but also in Delong’s tests when machine learning hyperparameters are re-tuned and optimized to each of the examined cohorts.

[0434] The same process was repeated across all other applicable models. In total, all of our newly established models vastly outperformed every single one of the non-invasive calculated indices, even following optimization, in all applicable cohorts during cross-validation and performed significantly better following hyperparameter tuning without cross-validation.

[0435] Methods Sample preparation

[0436] Global metabolomic profiling was performed by Metabolon Inc. (Morrisville, NC). Patient serum samples were ran blindly and consecutively, and robust batch, volume, and quality control sample corrections were implemented. Samples were maintained at -80°C until processed. Samples were prepared using the automated MICROLAB STAR® system from (Hamilton Company, Franklin, MA). Recovery standards were added prior to the first step in the extraction process for QC purposes. To remove protein, dissociate small molecules bound to protein or trapped in the precipitated protein matrix, and to recover chemically diverse metabolites, proteins were precipitated with methanol under vigorous shaking for 2 min using a GENOGRINDER® 2000 (GlenMills, Clifton, NJ) followed by centrifugation. Samples were placed briefly on a TURBOVAP® (Zymark) to remove the organic solvent. The sample extracts were stored overnight under nitrogen before preparation for analysis.

[0437] Quality assurance and control

[0438] Several types of controls were analyzed in concert with the experimental samples: a pooled matrix sample generated by taking a small volume of each experimental sample (or alternatively, use of a pool of well-characterized human plasma) served as a technical replicate throughout the data set; extracted water samples served as process blanks; and a cocktail of QC standards that were carefully chosen not to interfere with the measurement of endogenous compounds were spiked into every analyzed sample, allowed instrument performance monitoring and aided chromatographic alignment. Instrument variability was determined by calculating the median relative standard deviation (RSD) for the standards that were added to each sample prior to injection into the mass spectrometers. Overall process variability was determined by calculating the median RSD for all endogenous metabolites (i.e., noninstrument standards) present in 100% of the pooled matrix samples.

[0439] Ultrahigh Performance Liquid Chromatography-

[0440] Tandem Mass Spectroscopy

[0441] All methods utilized a Waters ACQUITY™ ultra-performance liquid chromatography (UPLC) and a Thermo Scientific Q-EXACTIVE™ high resolution / accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and ORBITRAP™ mass analyzer operated at 35,000 mass resolution. The dried sample extract was then reconstituted in solvents compatible to each of the four methods. Each reconstitution solvent contained a series of standards at fixed concentrations to ensure injection and chromatographic consistency. One aliquot was analyzed using acidic positive ion conditions, chromatographically optimized for more hydrophilic compounds (PosEarly). In this method, the extract was gradient eluted from a C18 column (Waters UPLC BEH C18-2.1 x100 mm, 1 .7 pm) using water and methanol, containing 0.05% perfluoropentanoic acid (PFPA) and 0.1 % formic acid. Another aliquot was also analyzed using acidic positive ion conditions, however it was chromatographically optimized for more hydrophobic compounds (PosLate). In this method, the extract was gradient eluted from the same aforementioned C18 column using methanol, acetonitrile, water, 0.05% PFPA and 0.01 % formic acid and was operated at an overall higher organic content. Another aliquot was analyzed using basic negative ion optimized conditions using a separate dedicated C18 column (Neg). The basic extracts were gradient eluted from the column using methanol and water, however with 6.5mM Ammonium Bicarbonate at pH 8. The fourth aliquot was analyzed via negative ionization following elution from a HILIC column (Waters UPLC BEH Amide 2.1x150 mm, 1 .7 pm) using a gradient consisting of water and acetonitrile with 10mM Ammonium Formate, pH 10.8 (HILIC). The MS analysis alternated between MS and data-dependent MSn scans using dynamic exclusion. The scan range varied slightly between methods but covered 70-1000 mass to charge ratio m / z. Raw data files are archived and extracted as described below.

[0442] Bioinformatics

[0443] The informatics system consisted of four major components: the Laboratory Information Management System (LIMS), the data extraction and peak-identification software, data processing tools for QC and compound identification, and a collection of information interpretation and visualization tools for use by data analysts. The hardware and software foundations for these informatics components were the LAN backbone, and a database server running Oracle 10.2.0.1 Enterprise Edition.

[0444] UMS

[0445] The purpose of the Metabolon LIMS system was to enable fully auditable laboratory automation through a secure, easy to use, and highly specialized system. The scope of the Metabolon LIMS system encompasses sample accessioning, sample preparation and instrumental analysis and reporting and advanced data analysis. All of the subsequent software systems are grounded in the LIMS data structures. It has been modified to leverage and interface with the in-house information extraction and data visualization systems, as well as third party instrumentation and data analysis software.

[0446] Data Extraction and Compound Identification

[0447] Raw data was extracted, peak-identified and QC processed using a combination of Metabolon developed software services. Each of these services perform a specific task independently, and they communicate / coordinate with each other using industry-standard protocols. Compounds were identified by comparison to library entries of purified standards or recurrent unknown entities. Metabolon maintains a library based on authenticated standards that contains the retention time / index (Rl), m / z, and fragmentation data on all molecules present in the library. Furthermore, biochemical identifications are based on three criteria: retention index within a narrow Rl window of the proposed identification, accurate mass match to the library + / - 10 parts per million (ppm), and the MS / MS forward and reverse scores between the experimental data and authentic standards. The MS / MS scores are based on a comparison of the ions present in the experimental spectrum to the ions present in the library spectrum. While there may be similarities between molecules based on one of these factors, the use of all three data points is utilized to distinguish and differentiate biochemicals. More than 5,400 commercially available purified or in-house synthesized standard compounds have been acquired and analyzed on all platforms for determination of their analytical characteristics. An additional 7000 mass spectral entries have been created for structurally unnamed biochemicals, which have been identified by virtue of their recurrent nature (both chromatographic and mass spectral). These compounds have the potential to be identified by future acquisition of a matching purified standard or by classical structural analysis. Metabolon continuously adds biologically-relevant compounds to its chemical library to further enhance its level of Tier 1 metabolite identifications.

[0448] Compound Quality Control A variety of curation procedures were carried out to ensure that a high-quality data set was made available for statistical analysis and data interpretation. The QC and curation processes were designed to ensure accurate and consistent identification of true chemical entities, and to remove or correct those representing system artifacts, mis-assignments, mis-integration and background noise. Metabolon data analysts use proprietary visualization and interpretation software to confirm the consistency of peak identification and integration among the various samples.

[0449] Metabolite Quantification and Data Normalization

[0450] Peaks were quantified using area-under-the-curve. For studies spanning multiple days, a data normalization step was performed to correct variation resulting from instrument inter-day tuning differences. Essentially, each compound was corrected in run-day blocks by registering the medians to equal one (1 .00) and normalizing each data point proportionately (termed the “block correction”). For studies that did not require more than one day of analysis, no normalization is necessary, other than for purposes of data visualization. In certain instances, biochemical data may have been normalized to an additional factor (e.g., cell counts, total protein as determined by Bradford assay, osmolality, etc.) to account for differences in metabolite levels due to differences in the amount of material present in each sample.

[0451] Clearing the metabolomics dataset

[0452] For the metabolomics dataset, metabolites with more than 30% missing values were excluded from the analysis. All other missing values were imputed using left-sided imputation (imputing with the minimum valid value measured within each metabolite). Unknown metabolites were included in the analysis and model building, and wherever applicable, models with unknown metabolites are presented alongside similar models excluding them. For the clinical - biochemical dataset, variables were missing completely at random per little’s MCAR test, and no clinical or hormonal variable ever displayed >10% missingness.

[0453] Complex lipid panel

[0454] Lipids were extracted from samples via a modified version of the extraction described by Matyash et al., J. Lipid Res., 49(5):1137-46, 2008, using methyl-tert-butyl ether in the presence of deuterated internal standards. The extracts were concentrated under nitrogen and reconstituted in 0.25 mL of 10mM ammonium acetate dichloromethane:methanol (50:50). The extracts were transferred to inserts and placed in vials for infusion-MS analysis, performed on a SHIMAZDU™ LC with nano PEEK tubing and the Sciex Selexlon-5500 QTRAP®. The samples were analyzed via both positive and negative mode electrospray. The 5500 QTRAP® scan was performed in MRM mode with the total of more than 1 ,100 MRMs. Individual lipid species were quantified by taking the peak area ratios of target compounds and their assigned internal standards, then multiplying by the concentration of internal standard added to the sample. Lipid species concentrations were background-subtracted using the concentrations detected in process blanks (water extracts) and run day normalized (when applicable). The resulting background- subtracted, run-day normalized lipid species concentrations were then used to calculate the lipid class and fatty acid total concentrations, as well as the mol% composition values for lipid species, lipid classes, and fatty acids. Clearing the lipidomics dataset

[0455] For the lipidomics dataset, missingness across lipid species was non-existent, and missingness across participants was minimal (<3%), with missing values being imputed using k-nearest neighbors. Lipid species concentrations were Iog2 transformed and analyzed as such. To assess lipids according to fatty acid type and perform enrichment analyses, the secondary fatty acid (or the only one annotated, in the case of triglycerides) was used to extract metadata on length, saturation, number of carbons and co3 / w6 status.

[0456] Statistical analysis - General

[0457] Statistical analyses were performed on SPSS® Statistics v. 25 (IBM®), PERSEUS™ (v. 1 .6.15.0), and Rstudio 2023.12.0 (Posit) using R v. 4.3.1 (The R Foundation). All visualizations were produced on Rstudio and MetaboAnalystR (v. 6.0). All reported p-values are two-tailed. Individual data points and statistical significance are all shown in the figures wherever applicable.

[0458] Volcano plots and enrichment analyses

[0459] To extract metabolomic and lipidomic snapshots across the six pre-determined histological thresholds, we plotted volcano plots of the significant Welch’s unpaired t-tests across all applicable molecules. We calculated fold-changes as the mean differences in Iog2 transformed concentrations (Iog2- concentration of people having the outcome, minus people not having the outcome) and compared by Welch’s t-tests. We applied the BH-FDR correction at a threshold of 0.05, and further set an absolute threshold of 1 .25 Iog2 fold-change for visualization purposes. For metabolomics, owing to the abundance of metabolite sub-pathways, we display increased vs decreased metabolites in red and blue, respectively, emphasizing the dots that retain significance after the BH-FDR correction. For lipids, we show the differences in main lipid classes and provide clear visual evidence of the lipidomic perturbations associated with each outcome. The data was subset using R packages dplyr (v1 .1 .3) and reshape2 (vO.8.9), and the tests and BH-FDR correction were ran using the built-in package stats (v4.3.1 ). Volcano plots were plotted with ggplot2 (v3.4.4) with annotations added by ggrepel (vO.9.3), and implemented a custom script to indicate proteins according to both raw significant and BH-FDR- corrected significant p- values, to capture the entire vector of changes.

[0460] Metabolite pathways were curated by Metabolon and were based on a combination of KEGG and HMDB insights, PubChem and domain-knowledge / literature, and structural homology. Lipids on the other hand were manually configured according to the length, saturation, carbon number and type of the secondary or listed / emphasized fatty acid within each lipid formula. Across all applicable categories and comparisons, Fisher’s exact tests were ran in contingency tables that only included BH-FDR significant variables, calculating fold-enrichment and p-value and indicating the cumulative fold-change across features and outcomes. Significant pathways, determined by their adjusted p-values, were visualized using bar plots with staggered, individual data points, to depict the distribution of fold changes among significant metabolites within each pathway. This visualization facilitated a comparative analysis across pathways, highlighting those with the most pronounced alterations.

[0461] Analysis of variance and complex heatmap We performed a comprehensive analysis of metabolites and lipids across the composite NAS score and the fibrosis score using packages RColorBrewer (v1 .1 .3), ComplexHeatmap (v2.15.4), dynamicTreeCut (v1 .63.1 ), tidyr (v1 .3.1 ), ggplot2 (v3.5.0), circlize (vO.4.16), and dplyr (v1 .1 .4). We first ran one-way ANOVA of all applicable variables across each score, correcting per BH-FDR. The pooled p- values across each score were then pooled. Initially, the top 100 variables for each score were selected based on the lowest adjusted p-values, with selection criteria being iteratively adjusted by increasing the number of top variables from each scoring system by increments of 10, until the desirable number of common variables (100 for metabolomics and 50 for lipidomics) were identified across both scoring systems.

[0462] Log2 concentrations were subsequently z-scored and grouped across categories of the fibrosis and NAS scores. Because very few people had NAS scores over 7, these were grouped in one designation. The same process was followed for clinical indices. The heatmap was hierarchically clustered, with complexHeatmap being configured to cut the map in three slices based on overall distributions across the scores, and was further annotated with tracks indicating variable type and differential regulation (by BH-FDR-corrected p-value) in several binary outcomes following a process as outlined in the volcano plots previously.

[0463] Heatmap clusters with similar patterns across the variable categories were then grouped together and plotted across the progression of each histological score. While lipid and clinical features were readily apparent, we plotted simple bar charts indicating the count of metabolic sub-pathways implicated in intracluster metabolites to facilitate cluster comprehension.

[0464] Dimensionality reduction

[0465] PCA biplots were drawn using significant omics and applicable clinical and biochemical features after Iog2 transformation and scaling. If any, missing values in clinical features were imputed according to column means. Plotted vectors were selected from the top loadings' absolute length and plotted as overlaid arrows over the biplots. PLS-DA in the Long-term study was performed on MetaboAnalystR which additionally extracted the top features and corresponding VI F values, reflecting the contribution of each variable to component separation.

[0466] Intra-metabolite correlations and WCNA

[0467] To fully delineate co-regulated metabolites and lipids under each binary disease state, we performed network correlation analyses implementing packages car (v. 3.1.2), rstatix (v. 0.7.2), readr (v. 2.1 .4), dplyr (v. 1.1.3), ggplot2 (v. 3.4.4), tidyr (v. 1 .3.0), Matrix (v. 1.6.3), igraph (v. 1 .5.1 ), ggraph (v. 2.1 .0), Ime4 (v. 1 .1 .35.1 ), ImerTest (v. 3.1 .3), circlize (v. 0.4.15), cluster Prof iler (v. 4.8.3), and enrichplot (v. 1 .20.3). We computed a Spearman correlation matrix and generated a corresponding p-value matrix, which was converted into a vector, adjusted per BH-FDR, and reconfigured back into a matrix. This produced a large correlation heatmap that was further refined by hierarchical clustering (Ward) and dynamic tree cutting set to follow the default unsupervised hierarchical clustering. To facilitate visualization and readability, we opted to further filter the heatmap, keeping only BH-FDR-significant correlations and applying a stringent threshold by only keeping absolute coefficients of >0.6 for metabolites, and >0.9 for lipids. We subsequently filtered each of the unsupervised clusters created by the heatmap accordingly and created protein-protein and protein-feature interaction / association networks. These cluster-specific networks were plotted using package igraph (v. 1 .5.1 ) and visualize with ggraph (v. 2.1 .0) using "graphopt" as a layout algorithm. We further implemented community detection using the WalkTrap method, to explore modular structures of coregulated molecules. We further emphasized interaction strength and directionality as defined by the value and direction of the correlation coefficients (red = positive, blue = negative, thicker and more opaque = higher in absolute value). The size of each node was proportional to the square root of the number of connections (degrees).

[0468] Within each of the identified metabolite clusters, we subsequently conducted enrichment analyses to identify overrepresented biological processes, applying the BH correction and calculating fold-enrichment to quantify pathway enrichment magnitude. Top pathways were selected and shown according to fold-enrichment. To fully visualize the entire network, we implemented a circos plot that acted as a more concise representation of the extended correlation heatmap, by showing only variables that held the most powerful and FDR-significant correlations, and thus facilitating a sector-specific representation of data, allowing for the clear differentiation and examination of connections within- and between-clusters. After re-configuring feature order using intra-cluster hierarchical clustering, we combined the circos plot showing correlations by adding tracks with circular dot plots of our GO enrichment analyses, and custom tracks for cluster-specific heatmaps showing the scaled relative Iog2 transformed levels of each feature at the individual level, plotting it as polygons instead of a regular track to account for unequally-sized node widths. To depict intra- and inter-cluster interactions, we drew links between sectors in the Circos plot, with colors and widths indicating correlation strength and statistical significance, mimicking the networks.

[0469] We further supplemented this information with network analysis of clinical outcomes utilizing weighted correlation network analysis (WGCNA). We generated similarity matrices from our data, which were transformed Topological Overlap Matrices (TOM) to accentuate the robust connections. Hierarchical clustering based on TOM identifies modules of highly correlated metabolites. Module-Trait Relationships: The eigengenes (principal components) representing each module's metabolic profile were correlated with clinical variables to identify modules significantly associated with traits of interest. The significance of these associations was visualized through a heatmap, overlaying p-values to highlight statistically significant correlations For modules of interest, metabolites were further examined, and modules underwent enrichment analyses in a manner similar to the one described above. This analysis was replicated with lipids, wherein module enrichment analysis focused on describing the structure, length, and saturation of within-module lipids.

[0470] Machine learning framework

[0471] Our machine learning pipeline was ran exclusively on Rstudio and implemented packages dplyr (v1 .1 .4), catboost (v1 .2.2), nnet (v7.3.19), caret (v6.0.94), pROC (v1 .18.5), ggplot2 (v3.5.0), ropls (v1 .26.4), randomForest (v4.7.1 .1 ), MASS (v7.3.60.0.1 ), missForest (v1 .5), broom (v1 .0.5), writexl (v1 .5.0), Rtsne (v0.17), ggrepel (vO.9.5), svglite (v2.1 .3), glmnet (v4.1 .8), utils (v4.1 .2), mixOmics (v6.18.1 ), doParallel (v1 .0.17), foreach (v1 .5.2), combinat (v0.0.8), RColorBrewer (v1 .1 .3), sdamr (v0.2.0), introdataviz (vO.0.0.9003), smotefamily (v1 .3.1 ), officer (vO.6.5), rsvg (v2.6.0), rmarkdown (v2.26), magrittr (v2.0.3), rvg (vO.3.3), ggpubr (v0.6.0), tidyr (v1 .3.1 ), gridExtra (v2.3), patchwork (v1 .2.0), grid (v4.1 .2), and ggdendro (v0.2.0). For the predictive models, we implemented Categorical Gradient Boosting Machines (CatBoost). CatBoost is a gradient-boosting decision tree-based machine learning algorithm that specializes in handling and naturally processing categorical variables (akin to our binary outcomes and several of our clinical predictors) without the need for extensive pre-processing, whereas it can also function independently of missing values, data distribution, or variable nature (i.e. factor, integer, numeric), and is robust without excessive hyperparameter tuning. The loss function was set as LogLoss, suitable for binary classification, and the internal model bootstrap type was set as Minimum Variance Sampling. We set initial model architecture using conservative settings, implementing a tree depth of 6, a learning rate of 0.05 (default learning rate for our dataset was set at 0.053 by CatBoost), and an L2 regularization of 1 . We further configured model convergence at 200 iterations to avoid overtraining and further applied an overtraining detector to halt model training at 20 iterations upon detection of performance plateaus.

[0472] Feature selection

[0473] To narrow down clinical, biochemical, and omics variables and select the best and most relevant features for each predictive model, we implemented, exclusively within the training cohort, 10-fold crossvalidated recursive feature elimination (FtFE) with all clinical and metabolomic variables for each target histological outcome, using a series of CatBoost-specific functions written for R package “caret” that separately ranked all applicable variables per model importance score. Because of computational intensity and CatBoost function complexity, variable scores were consistent but directly dependent on the number and nature of all co-variables subjected to RFE. Hence to confirm feature importance accuracy and fully encompass training data dimensionality and variable contribution to outcome separation, we confirmed RFE results by running a principal component analysis (PCA) of the top 100 variables selected via RFE, and an elastic net logistic regression of these top 100 variables with the target outcome. Variables were then sorted according to an aggregate final score, and the top 5 variables (or up to 20, with manual exclusion of redundant features) were selected as final candidates to be screened during model building. Catboost, like most supervised machine learning decision tree models, is by nature immune to by multicollinearity, however a reasonable approach was to exclude models with overly redundant variables (i.e. existence of diabetes alongside Insulinemia, HOMA-IR alongside insulinemia, or Total MetS components alongside absolute waist circumference; but not, for instance Total MetS components alongside the existence of diabetes, since IGT does not necessarily incorporate diabetes).

[0474] Model building and training

[0475] For each target histological outcome, we initially set 1 -3 predetermined variables based on domain knowledge and / or correlation coefficients based on our prior analysis, and set them as “locked”, predetermined model features. We then implemented two approaches: a) A conservative approach selecting the top 1 -2 performing metabolite(s) or lipid(s) from our feature selection process (conservative models - green curves); b) A floating sequential feature selection algorithm, with a forward and a backward step, that selected features from the top variables according to their combined effect on model improvement (floating models - blue curves). The algorithm started from the predetermined features and selected top-performing variables from the set, adding (forward step) or removing them (backward step) from the model accordingly, guided by the sequential improvement in model AUROC in the validation cohort. This iterative process continued until additional variables yielded negligible improvement or even decreased model AUROC. This approach ensured a balance between model complexity and accuracy, allowing for the construction of robust models with adequate and clinically relevant features, and aiming to minimize the risk of overfitting that could arise from indiscriminately including a large number of variables based solely on their RFE importance rank. We present both the aggregate final models and the progressive AUROC curves denoting model building process, benchmarked on the validation cohort. Model hyperparameters were tuned following feature selection via a nested cross-validated grid search, wherein the training cohort was split into two folds, each one 5-fold cross-validated in itself, until the optimal hyperparameters (iterations, depth, learning rate, 12 regularization) based on frequency across all folds were selected. To gauge overtraining, we calculated intra-model SHAP values and plotted learning curves across the training and validation sets, progressing at incremental slices of 5 or 10% depending on the distribution of the binary feature of interest.

[0476] Model benchmarking and evaluation

[0477] The trained models were saved and applied to the Training, Validation, and Entire (Training + Validation) cohorts using 10- or 5-fold cross-validation with 10x random resampling (10-fold in training or entire, 5-fold in validation due to sample size). Folds were created at random to ensure unbiased distribution of the primary outcome, and each evaluation calculated the AUROC and optimal Youden’s index, thus additionally calculating optimized accuracy, PPV, NPV, sensitivity and specificity, which were eventually summarized, in order to fully capture model behavior in each dataset and slice of data. We further performed a series of sensitivity analyses to fully assess floating model generalizability, precision, and robustness. To ensure non-confounding by clinical or demographic variables relevant to MASLD, we further evaluated model performance in clinically relevant subgroups of the entire cohort, performing subgroup sensitivity analyses according to MASLD status whenever applicable, obesity, diabetes, type of referral center, and participant home country. To fully capture model performance at the individual level, we performed leave-one-out cross-validation within the entire cohort, essentially applying the model to each individual independently, and calculating aggregate model AUG, accuracy, and sensitivity and specificity values at the optimal machine learning prediction probability cutoff. Even though an early overtraining detector was the key component in our model architecture, we further performed learning curve analysis to comprehensively gauge model performance and potential overtraining (i.e. model performing gradually worse in validation cohort with the addition of more data). To that end, we plotted model AUROC against the incremental proportion of data used to train and validate the model in both the training and validation cohorts (starting from 10% of total respective data and increasing in steps of 10%) repeated 20 times with randomized data reshuffling. We further computed patient-specific Shapley values for each variable within our models. Shapley values quantify the marginal contribution of each feature to the predictive probability of the machine learning model (i.e. histological outcomes, in our case) for an individual patient, considering all possible combinations of other features. It is important to note that all of our predictive models function as binary classifiers, wherein 0 denotes control participants (without the outcome) and 1 indicates participants with the outcome; hence in interpreting machine learning probabilities, including Shapley values, a positive probability would indicate a right-sided model prediction (presence of the outcome), whereas a negative probability would indicate a left-sided prediction (absence of the outcome). Comparison of model metrics with pre-established indexes

[0478] To fully elucidate model performance and establish a case for superiority over other non-invasive indices of MASLD stratification, we compared model performances with fifteen calculated indices (Table 7). These indices, representing pre-calculated scores, were re-optimized across every comparison regardless of their intended designed outcome, to fully ensure the robustness of our new models. The comparisons were effectuated in 5-fold cross-validated, 10x resampled iterations of a) the validation cohort; b) the entire dataset; c) a random 90-person slice of the entire dataset to match the size of the validation cohort. We plotted results of all applicable metrics as individual data points on staggered violin plots, and performed unpaired Welch’s t-tests to compare all AUROCs. We further performed the evaluations without cross-val...

Claims

1. CLAIMS1 . A method for diagnosing metabolic dysfunction-associated steatotic liver disease (MASLD) in a subject comprising determining a change in a level of at least one biomarker in a biological sample obtained from a subject relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 2-aminoadipate, 4-hydroxyphenylacetylglutamine, Alanine aminotransferase (ALT), arabitol / xylitol, C-glycosyltryptophan, DAG(14:0 / 20:0), DAG(16:0 / 18:1 ), DAG(16:0 / 18:3), DAG(18:0 / 18:1 ), DAG (18:2 / 18:3), DCER(22:0), DCER(26:0), dimethylarginine (SDMA + ADMA), glucose, glutamate, MAG(18:1 ), mannonate, mannose, N6-carbamoylthreonyladenosine, N-acetyltryptophan, pantothenate, ribitol, SM(22:0), sphingadienine, sphinganine, sphingosine, TAG47:1 -FA16:1 , TAG50:1 -FA18:0, TAG51 :1 -FA17:0, TAG52:1 -FA16:0, TAG52:3-FA20:3, TAG54:2-FA16:0, TAG54:6-FA18:3, TAG54:6- FA20:5, TAG58:6-FA20:4, X-23654, androsterone glucuronide, CER(14:0), guanosine, PC(14:0 / 20:4), PC(15:0 / 18:2), pregnanediol-3-glucuronide, and TAG57:3-FA18:2.

2. The method of claim 1 , wherein:(a) an increased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 2- aminoadipate, 4-hydroxyphenylacetylglutamine, Alanine aminotransferase (ALT), arabitol / xylitol, C- glycosyltryptophan, DAG(14:0 / 20:0), DAG(16:0 / 18:1 ), DAG(16:0 / 18:3), DAG(18:0 / 18:1 ), DAG (18:2 / 18:3), DCER(22:0), DCER(26:0), dimethylarginine (SDMA + ADMA), glucose, glutamate, MAG(18:1 ), mannonate, mannose, N6-carbamoylthreonyladenosine, N-acetyltryptophan, pantothenate, ribitol, SM(22:0), sphingadienine, sphinganine, sphingosine, TAG47:1 -FA16:1 , TAG50:1 -FA18:0, TAG51 :1 - FA17:0, TAG52:1 -FA16:0, TAG52:3-FA20:3, TAG54:2-FA16:0, TAG54:6-FA18:3, TAG54:6-FA20:5, TAG58:6-FA20:4, and X-23654;and / or(b) a decreased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: androsterone glucuronide, CER(14:0), guanosine, PC(14:0 / 20:4), PC(15:0 / 18:2), pregnanediol-3- glucuronide, and TAG57:3-FA18:2.

3. A method for diagnosing metabolic dysfunction-associated steatotic liver disease (MASLD) in a subject comprising determining a change in a level of at least one biomarker in a biological sample obtained from a subject relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha- dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a- ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, X-26158, 2'-deoxyuridine, inosine, leucylalanine, N- acetylmethionine, and threonylphenylalanine.

4. The method of claim 3, wherein:(a) an increased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 1 - stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5- cholestenoate, 3-indoleg lyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and / or(b) a decreased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 2'- deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.

5. The method of claim 3 or 4, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

6. The method of any one of claims 3-5, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

7. The method of any one of claim 1 -6, wherein the subject has stage 1 liver fibrosis (F1 ), stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), or stage 4 liver fibrosis (F4).

8. The method of any one of claims 1 -7, wherein the MASLD is metabolic dysfunction-associated steatotic liver (MAFL) or metabolic dysfunction-associated steatohepatitis (MASH).

9. The method of claim 8, wherein the MASH is at-risk MASH.

10. The method of claim 9, wherein the at-risk MASH is MASH F2, MASH F3, MASH F4, MASH F2-F3, or MASH F3-F4.11 . The method of claim 10, wherein the at-risk MASH is MASH F2-F3.

12. The method of claim 10, wherein the at-risk MASH is MASH F3-F4.

13. The method of any one of claims 1 -12, wherein the subject has a non-alcoholic steatohepatitis activity (NAS) score >4.

14. The method of any one of claims 1 -13, wherein the method further comprises administering a MASLD therapy to the subject.

15. A method of treating MASLD in a subject comprising administering a MASLD therapy to the subject, wherein the subject is determined to have MASLD according to the method of any one of claims 1 -13.

16. A method of treating MASH in a subject comprising administering a MASLD therapy to the subject, wherein the subject is determined to have MASH according to the method of any one of claims 3-13.

17. A method of treating MASLD in a subject comprising:(a) identifying the subject as having MASLD by determining a change in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 2-aminoadipate, 4- hydroxyphenylacetylglutamine, Alanine aminotransferase (ALT), arabitol / xylitol, C-glycosyltryptophan, DAG(14:0 / 20:0), DAG(16:0 / 18:1 ), DAG(16:0 / 18:3), DAG(18:0 / 18:1 ), DAG(18:2 / 18:3), DCER(22:0), DCER(26:0), dimethylarginine (SDMA + ADMA), glucose, glutamate, MAG(18:1 ), mannonate, mannose, N6-carbamoylthreonyladenosine, N-acetyltryptophan, pantothenate, ribitol, SM(22:0), sphingadienine, sphinganine, sphingosine, TAG47:1 -FA16:1 , TAG50:1 -FA18:0, TAG51 :1 -FA17:0, TAG52:1 -FA16:0, TAG52:3-FA20:3, TAG54:2-FA16:0, TAG54:6-FA18:3, TAG54:6-FA20:5, TAG58:6-FA20:4, X-23654, androsterone glucuronide, CER(14:0), guanosine, PC(14:0 / 20:4), PC(15:0 / 18:2), pregnanediol-3- glucuronide, and TAG57:3-FA18:2; and(b) administering a MASLD therapy to the subject.

18. The method of claim 17, wherein:(a) an increased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 2- aminoadipate, 4-hydroxyphenylacetylglutamine, Alanine aminotransferase (ALT), arabitol / xylitol, C- glycosyltryptophan, DAG(14:0 / 20:0), DAG(16:0 / 18:1 ), DAG(16:0 / 18:3), DAG(18:0 / 18:1 ), DAG (18:2 / 18:3), DCER(22:0), DCER(26:0), dimethylarginine (SDMA + ADMA), glucose, glutamate, MAG(18:1 ), mannonate, mannose, N6-carbamoylthreonyladenosine, N-acetyltryptophan, pantothenate, ribitol, SM(22:0), sphingadienine, sphinganine, sphingosine, TAG47:1 -FA16:1 , TAG50:1 -FA18:0, TAG51 :1 - FA17:0, TAG52:1 -FA16:0, TAG52:3-FA20:3, TAG54:2-FA16:0, TAG54:6-FA18:3, TAG54:6-FA20:5, TAG58:6-FA20:4, and X-23654; and / or(b) a decreased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: androsterone glucuronide, CER(14:0), guanosine, PC(14:0 / 20:4), PC(15:0 / 18:2), pregnanediol-3- glucuronide, and TAG57:3-FA18:2.

19. A method of treating MASLD in a subject comprising:(a) identifying the subject as having MASLD by determining a change in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0),dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, X-26158, 2'- deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine; and(b) administering a MASLD therapy to the subject.

20. The method of claim 19, wherein:(a) an increased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 1 - stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5- cholestenoate, 3-indoleg lyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and / or(b) a decreased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 2'- deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.21 . The method of claim 19 or 20, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

22. The method of any one of claims 19-21 , wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

23. The method of any one of claims 19-22, wherein the MASLD is MAFL or MASH.

24. The method of any one of claims 17-23, wherein the subject has stage 1 liver fibrosis (F1 ), stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), or stage 4 liver fibrosis (F4).

25. The method of claim 23, wherein the at-risk MASH is MASH F2, MASH F3, MASH F4, MASH F2-F3, or MASH F3-F4.

26. The method of claim 25, wherein the at-risk MASH is MASH F2-F3.

27. The method of claim 25, wherein the at-risk MASH is MASH F3-F4.

28. The method of any one of claims 14-20, wherein the subject has a NAS score >4.

29. The method of any one of claims 1 -28, wherein the level of the at least one biomarker is increased or decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference.

30. The method of any one of claims 14-29, wherein the method further comprises a step of monitoring treatment efficacy of the MASLD therapy administered to the subject.31 . The method of claim 30, wherein the treatment efficacy is monitored once every 10 years, once every5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily.

32. The method of claims 30 or 31 , wherein the step of monitoring treatment efficacy comprises:(a) identifying the subject’s condition as:(i) progressive, wherein the level of the at least one biomarker in the biological sample obtained from the subject has increased over time relative to a reference level or a previous sample obtained from the subject; or(ii) stabilized, wherein the level of the at least one biomarker in a biological sample obtained from the subject has remained unchanged over time relative to a reference level or a previous sample obtained from the subject, and(b) modifying the MASLD therapy administered to the subject.

33. The method of claim 32, wherein modifying the MASLD therapy administered to the subject comprises increasing a dosage and / or frequency of administration of the MASLD therapy to the subject.34.The method of claim 33, wherein the dosage is increased by at least 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is increased to at least6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly.

35. The method of any one of claims 32-34, wherein the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

36. The method of claims 30 or 31 , wherein the step of monitoring treatment efficacy comprises:(a) identifying the subject’s condition as in remission, wherein the level of the at least one biomarker in the biological sample obtained from the subject has decreased over time relative to a reference level or a previous sample obtained from the subject, and(b) modifying the MASLD therapy administered to the subject.

37. The method of claim 36, wherein modifying the MASLD therapy administered to the subject comprises decreasing a dosage and / or frequency of administration of the MASLD therapy to the subject.

38. The method of claim 37, wherein the dosage is decreased by at least 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years.

39. The method of any one of claims 36-38, wherein the level of the at least one biomarker has decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

40. A method of modifying a MASLD therapy in a subject comprising:(a) increasing a dosage and / or frequency of administration of a MASLD therapy in the subject, wherein the subject is determined to have an increased or unchanged level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl- GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3- indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2),DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the modified MASLD therapy to the subject.41 . The method of claim 40, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

42. The method of claim 40 or 41 , wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

43. The method of any one of claims 40-42, wherein the dosage is increased by at least 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly.

44. The method of any one of claims 40-43, wherein the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

45. A method of modifying a MASLD therapy in a subject comprising:(a) decreasing a dosage and / or frequency of administration of a MASLD therapy in the subject, wherein the subject is determined to have a decreased level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3- indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2),DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the modified MASLD therapy to the subject.

46. The method of claim 45, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

47. The method of claim 45 or 46, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

48. The method of any one of claims 45-47, wherein the dosage is decreased by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%,120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years.

49. The method of any one of claims 45-48, wherein the level of the at least one biomarker has decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

50. A method of monitoring treatment efficacy in a subject being treated with a MASLD therapy comprising:(a) detecting an increase in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the MASLD therapy administered to the subject.51 . The method of claim 50, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

52. The method of claim 50 or 51 , wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

53. The method of any one of claims 50-52, wherein treatment efficacy is monitored once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily.

54. The method of any one of claims 50-53, wherein modifying the MASLD therapy administered to the subject comprises increasing a dosage and / or frequency of administration of the MASLD therapy to the subject.

55. The method of claim 54, wherein the dosage is increased by at least 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly.

56. The method of any one of claims 50-55, wherein the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

57. A method of monitoring treatment efficacy in a subject being treated with a MASLD therapy, the method comprising:(a) detecting a decrease in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the MASLD therapy administered to the subject.

58. The method of claim 57, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

59. The method of claim 57 or 58, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

60. The method of any one of claims 57-59, wherein treatment efficacy is monitored by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, biweekly, weekly, or daily.61 . The method of any one of claims 57-60, wherein modifying the MASLD therapy administered to the subject comprises decreasing a dosage and / or frequency of administration of the MASLD therapy to the subject.

62. The method of claim 61 , wherein the dosage is decreased by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years.

63. The method of any one of claims 57-62, wherein the level of the at least one biomarker has decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

64. The method of any one of claims 40-63, wherein the subject has MASLD.

65. The method of claim 64, wherein the MASLD is MAFL or MASH.

66. The method of claim 64 or 65, wherein the subject has stage 1 liver fibrosis (F1 ), stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3) , or stage 4 liver fibrosis (F4).

67. The method of claim 65, wherein the MASH is at-risk MASH.

68. The method of claim 67 wherein the at-risk MASH is MASH F2, MASH F3, MASH F4, MASH F2-F3, or MASH F3-F4.

69. The method of claim 68, wherein the at-risk MASH is MASH F2-F3.

70. The method of claim 68, wherein the at-risk MASH is MASH F3-F4.71 . The method of any one of claims 40-70, wherein the subject has a NAS score >4.

72. A method of monitoring disease progression in a subject with MASH comprising:(a) detecting a change in a level of at least one biomarker in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the change in the level of the at least one biomarker indicates that MASH has progressed to at-risk MASH; and(b) administering a MASLD therapy to said subject.

73. The method of claim 72, wherein:(a) an increased level of the at least one biomarker relative to a reference level indicates that MASH has progressed to at-risk MASH, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha- dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a- ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158;and / or(b) a decreased level of the at least one biomarker relative to a reference level indicates that MASH has progressed to at-risk MASH, wherein the at least one biomarker is selected from the group consisting of: 2'-deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.

74. The method of claim 72 or 73, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

75. The method of any one of claims 72-74, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

76. The method of any one of claims 72-75, wherein disease progression is monitored by evaluating the subject once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, biweekly, weekly, or daily.

77. The method of any one of claim 14-76, wherein the MASLD therapy is or comprises weight loss, a dietary change, increased physical activity, a bariatric surgery, an anti-inflammatory agent, an antiapoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a peroxisome proliferator-activated receptor (PPAR) modulator, a thyroid receptor beta agonist, a fibroblast growth factor 21 (FGF21 ) agonist, a glucagon-like peptide-1 (GLP-1 ) agonist, an incretin receptor agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an antihypertensive drug, a farnesoid X receptor (FXR) agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), and / or JKB-121 .

78. The method of claim 77 wherein the thyroid beta receptor agonist is resmetirom (REZDIFFRA™).

79. The method of any one of claims 14-78, wherein the method reduces or delays the progression of MASLD.

80. The method of claim 79, wherein the method reduces or delays the progression of MASLD by at least 1 day, 5 days, 10 days, 20 days, 1 month, 2 months, 3 months, 4 months, 5 months, 6 months, 7 months 8 months, 9 months, 10 months, 11 months, 1 year, 1 .5 years, 2 years, 2.5 years, 3 years, 3.5 years, 4 years, 5 years, 6 years, 7 years, 8 years, 9 years, 10 years, 15 years, 20 years, 25 years, 30 years, 35 years, 40 years, 50 years, 55 years, or more.81 . A method for diagnosing liver fibrosis in a subject comprising determining a level of at least one biomarker in a biological sample obtained from the subject, wherein:(a) an increased level of the at least one biomarker relative to a reference is indicative of the presence of liver fibrosis, wherein the at least one biomarker is selected from the group consisting of: 1 - stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5- cholestenoate, 3-indoleg lyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate,alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and / or(b) a decreased level of the at least one biomarker relative to a reference is indicative of the presence of liver fibrosis, wherein the at least one biomarker is selected from the group consisting of: 2'- deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.

82. The method of claim 81 , wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

83. The method of claim 81 or 82, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

84. The method of any one of claims 81 -83, wherein the liver fibrosis is stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), stage 4 liver fibrosis (F4), stage F2-F3 liver fibrosis, or stage F3-F4 liver fibrosis.

85. The method of any one of claims 81 -84, wherein the liver fibrosis is F2-F3.

86. The method of any one of claims 81 -84, wherein the liver fibrosis is F3-F4.

87. The method of any one of claims 81 -84, wherein the method further comprises administering a liver fibrosis therapy to the subject.

88. A method of treating liver fibrosis in a subject comprising administering a liver fibrosis therapy to the subject, wherein the subject is determined to have liver fibrosis according to the method of any one of claims 81 -86.

89. A method of treating liver fibrosis in a subject, the method comprising:(a) identifying the subject as having liver fibrosis by determining that a level of at least one biomarker in a biological sample obtained from the subject is changed relative to a reference level; and(b) administering a liver fibrosis therapy to the subject.

90. The method of claim 89, wherein:(a) an increased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 1 - stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5- cholestenoate, 3-indoleg lyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate,alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and / or(b) a decreased level of the at least one biomarker relative to a reference level is indicative of the presence of MASLD, wherein the at least one biomarker is selected from the group consisting of: 2'- deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.91 . The method of claim 90, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

92. The method of any one of claims 89-91 , wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

93. The method of any one of claim 89-92, wherein the liver fibrosis is stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), stage 4 liver fibrosis (F4), stage F2-F3 liver fibrosis, or stage F3-F4 liver fibrosis.

94. The method of any one of claims 89-93 wherein the liver fibrosis is F2-F3.

95. The method of any one of claims 89-93, wherein the liver fibrosis is F3-F4.

96. The method of any one of claims 81 -95, wherein the subject has a NAS score >4.

97. The method of any one of claims 81 -96, wherein the level of the at least one biomarker is increased or decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level.

98. The method of any one of claims 81 -97, wherein the method further comprises a step of monitoring treatment efficacy of the liver fibrosis therapy administered to the subject.

99. The method of claim 98, wherein treatment efficacy is monitored once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily.

100. The method of claim 98 or 99, wherein the step of monitoring treatment efficacy comprises:(a) identifying the subject’s condition as:(i) progressive, wherein the level of the at least one biomarker in a biological sample obtained from the subject has increased over time relative to a reference level or a previous sample obtained from the subject; or(ii) stabilized, wherein the level of the at least one biomarker in a biological sample obtained from the subject has remained unchanged over time relative to a reference level or a previous sample obtained from the subject; and(b) modifying the liver fibrosis therapy administered to the subject.101 . The method of claim 100, wherein modifying the liver fibrosis therapy administered to the subject comprises increasing a dosage and / or frequency of administration of the liver fibrosis therapy to the subject.

102. The method of claim 101 , wherein the dosage is increased by at least 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly.

103. The method of any one of claims 100-102, wherein the level of the at least one biomarker has increased or decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

104. The method of claim 98 or 99, wherein the step of monitoring treatment efficacy comprises:(a) identifying the subject’s condition as in remission, wherein the level of the at least one biomarker in the biological sample obtained from the subject has decreased over time relative to a reference level or a previous sample obtained from the subject; and(b) modifying the liver fibrosis therapy administered to the subject.

105. The method of claim 104, wherein modifying the liver fibrosis therapy administered to the subject comprises decreasing a dosage and / or frequency of administration of the liver fibrosis therapy to the subject.

106. The method of claim 105, wherein the dosage is decreased by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years.

107. The method of any one of claims 104-106, wherein the level of the at least one biomarker has decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

108. A method of modifying a liver fibrosis therapy in a subject comprising:(a) increasing a dosage and / or frequency of administration of a liver fibrosis therapy in the subject, wherein the subject is determined to have an increased or unchanged level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha- dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a- ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the modified liver fibrosis therapy to the subject.

109. The method of claim 108, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

110. The method of claim 108 or 109, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

111. The method of any one of claims 108-110, wherein the dosage is increased by at least 5%, 10%,15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or wherein the frequency of administration is increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly.

112. The method of any one of claims 108-111 , wherein the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

113. A method of modifying a liver fibrosis therapy in a subject comprising:(a) decreasing a dosage and / or frequency of administration of a liver fibrosis therapy in the subject, wherein the subject is determined to have an decreased level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl- GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3- indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2),DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the modified liver fibrosis therapy to the subject.

114. The method of claim 113, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

115. The method of claim 113 or 114, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

116. The method of any one of claims 113-115, wherein the dosage is decreased by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%,120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years.

117. The method of any one of claims 113-116, wherein the level of the at least one biomarker has decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

118. A method of monitoring treatment efficacy in a subject being treated with a liver fibrosis therapy, the method comprising:(a) detecting an increase in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0),dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the liver fibrosis therapy administered to the subject,119. The method of claim 118, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

120. The method of claim 118 or 119, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.121 . The method of any one of claims 118-120, wherein treatment efficacy is monitored once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily.

122. The method of any one of claims 118-121 , wherein modifying the liver fibrosis therapy administered to the subject comprises increasing a dosage and / or frequency of administration of the MASLD therapy to the subject.

123. The method of any one of claims 118-122, wherein the dosage is increased by at least 5%, 10%,15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is increased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, or yearly.

124. The method of any one of claims 118-123, wherein the level of the at least one biomarker has increased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

125. A method of monitoring treatment efficacy in a subject being treated with a liver fibrosis therapy, the method comprising:(a) detecting a decrease in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the liver fibrosis therapy administered to the subject,126. The method of claim 125, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4.

127. The method of claim 125 or 126, wherein the at least one biomarker is selected from the group consisting of: 3-ureidopropionate, kynurenine, and a-ketoglutarate.

128. The method of any one of claims 125-127, wherein treatment efficacy is monitored once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily.

129. The method of any one of claims 125-128, wherein modifying the liver fibrosis therapy administered to the subject comprises decreasing a dosage and / or frequency of administration of the liver fibrosis therapy to the subject.

130. The method of claim 129, wherein the dosage is decreased by 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 150%, 160%, 170%, 180%, 190%, 200%, or more, and / or the frequency of administration is decreased to at least 6 times daily, 5 times daily, 4 times daily, 3 times daily, 2 times daily, once daily, every other day, two times weekly, once weekly, biweekly, monthly, bimonthly, yearly, once every 2 years, once every 3 years, once every 4 years, once every 5 years, or once every 10 years.131 . The method of any one of claims 125-130, wherein the level of the at least one biomarker has decreased by at least 5%, 7%, 10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 100%, 110%, 120%, 130%, 140%, 150%, 160%, 170%, 180%, 190%, 200%, 220%, 240%, 250%, 260%, 280%, 300%, or more relative to the reference level or the previous sample obtained from the subject.

132. The method of claims 108-131 , wherein the subject has liver fibrosis.

133. The method of claim 132, wherein the liver fibrosis is stage 2 liver fibrosis (F2), stage 3 liver fibrosis (F3), stage 4 liver fibrosis (F4) , stage F2-F3 liver fibrosis, or stage F3-F4 liver fibrosis.

134. The method of claim 132 or 133, wherein the liver fibrosis is F2-F3.

135. The method of claim 132 or 133, wherein the liver fibrosis is F3-F4.

136. The method of any one of claims 87-135, wherein the method reduces or delays the progression fibrosis.

137. The method of claim 136, wherein the method reduces or delays the progression of fibrosis by at least 1 day, 5 days, 10 days, 20 days, 1 month, 2 months, 3 months, 4 months, 5 months, 6 months, 7 months 8 months, 9 months, 10 months, 11 months, 1 year, 1 .5 years, 2 years, 2.5 years, 3 years, 3.5 years, 4 years, 5 years, 6 years, 7 years, 8 years, 9 years, 10 years, 15 years, 20 years, 25 years, 30 years, 35 years, 40 years, 50 years, 55 years, or more.

138. A method of monitoring disease progression in a subject with liver fibrosis, the method comprising:(a) detecting an increase in a level of at least one biomarker in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the increase in the level of the at least one biomarker indicates that the liver fibrosis has progressed to stage F2, F3, or F4 liver fibrosis, and wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3- indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2),DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering a liver fibrosis therapy to said subject.

139. The method of claim 138, wherein the disease progression is monitored once every 10 years, once every 5 years, once every 4 years, once every 3 years, once every 2 years, once yearly, twice a year, three times a year, four times a year, six times a year, monthly, bi-weekly, weekly, or daily.

140. The method of claim 139, wherein the liver fibrosis has progressed from F2 to F3.141 . The method of claim 139, wherein the liver fibrosis has progressed from F3 to F4 .

142. The method of any one of any one of claims 87-141 , wherein the liver fibrosis therapy is or comprises weight loss, a dietary change, increased physical activity, a bariatric surgery, a liver transplant an ani-fibrotic agent, an anti-inflammatory agent, an anti-apoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a PPAR agonist, a thyroid receptor beta agonist, a fibroblast growth factor 21 (FGF21 ) agonist, a glucagon-like peptide-1 (GLP-1 ) agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a FXR agonist, MGL-3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, CVC, or JKB-121 .

143. The method of claim 142, wherein the thyroid beta receptor agonist is resmetirom (REZDIFFRA™).

144. The method of any one of claims 1 -143, wherein the method further comprises determining one or more additional parameters selected from the group consisting of: a liver transaminase level, a platelet count, an albumin level, age, a body mass index (BMI), and a patatin-like phospholipase domaincontaining protein 3 (PNPLA3) genotype.

145. The method of claim 144, wherein the liver transaminase comprises alanine aminotransferase (ALT) and / or aspartate aminotransferase (AST).

146. The method of claim 145, wherein the subject comprises:(a) an AST level above 36 international units per liter (IU / L) or below 10 IU / L;(b) an ALT level above 56 IU / L or below 7 IU / L;(c) a BMI above 24.9 kilograms per square meter (kg / m2) or below 18.5 kg / m2;(d) a total number of metabolic syndrome (TotalMetS) above 3;(f) a platelet count above 450 x 109per liter (L) or below 150 x 109 / L; and / or(g) a hemoglobin A1 (HbA1c) greater than or equal to 5.7%147. The method of claim 145, wherein the subject comprises:(a) an AST level above 107 IU / L or below 9 IU / L;(b) an ALT level above 168 IU / L or below 5 IU / L;(c) a BMI above 62 kilograms per square meter kg / m2or below 19.6 kg / m2;(d) a TotalMetS above 3;(f) a platelet count above 448 x 109 / L or below 54 x 109 / L; and / or(g) a hemoglobin A1 (HbA1 c) greater than 8.4% or below 4.3%.

148. A method of stratifying a subject having MASLD for treatment comprising:(a) determining a change in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level; and(b) classifying the status of the subject’s MASLD as at-risk MASH based on the change determined in step (a).

149. The method of claim 148, wherein(a) an increased level of the at least one biomarker relative to the reference level classifies the status of the subject’s MASLD as at-risk MASH, wherein the at least one biomarker selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha- dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a- ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and / or(b) a decreased level of the at least one biomarker relative to the reference level classifies the status of the subject’s MASLD as at-risk MASH, wherein the at least one biomarker selected from the group consisting of: 2'-deoxyuridine, inosine, leucylalanine, N-acetylmethionine, and threonylphenylalanine.

150. The method of claim 148 or 149, further comprising classifying the subject as a candidate or noncandidate for receiving a MASLD therapy based on the status of the subject’s MASLD, wherein:(a) the subject’s MASLD is classified as at-risk MASH, thereby classifying the subject as a candidate to receive at least one MASLD therapy selected from the group consisting of: an antiinflammatory agent, an anti-apoptosis agent, a medication to treat insulin resistance or type 2 diabetes, a medication to improve cholesterol, Vitamin E, a peroxisome proliferator-activated receptor (PPAR) modulator, a thyroid receptor beta agonist, an incretin receptor agonist, glutathione, usrodeoxycholic acid, pemafibrate, aramchol, GS0976, an anti-hypertensive drug, a farnesoid X receptor (FXR) agonist, MGL- 3196, Bl 1467335, IMM-124E, solithromycin, BMS-986036, Cenicriviroc (CVC), JKB-121 , and resmetirom (REZDIFFRA™); or(b) the subject’s MASLD is not classified as at-risk MASH, thereby classifying the subject as a non-candidate for resmetirom (REZDIFFRA™).151 . The method of any one of claims 1 -150, wherein the reference level is established by performing classifier training using 3-ureidopropionate, kynurenine, a-ketoglutarate, mannonate, TAG54:4 / FA22:4, X- 26158, ALTUIL, 2’-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2- methylcitrate / homocitrate, SAH, and 2'-O-methyluridine levels obtained from subjects clinically diagnosed as having MASH, subjects clinically diagnosed as having fibrosis, and healthy subjects.

152. The method of one of claims 1 -151 , wherein the reference level is established by training a machine learning algorithm using a reference dataset.

153. The method of any one of claims 1 -152, wherein the level of the at least one biomarker is determined by one or more of mass spectrometry (MS), liquid chromatography (LC)-MS (LC / MS), liquidchromatography tandem-mass-spectrometry (LC-MS / MS), gas chromatography (GC), or enzyme linked immunosorbent assay (ELISA).

154. The method of any one of claims 1 -153, wherein the at least one biomarker comprises at least two biomarkers, at least three biomarkers, at least four biomarkers, at least five biomarkers, or more.

155. The method of claim 154, wherein:(a) the at least one biomarker comprises:(i) 3-ureidopropionate;(ii) kynurenine;(iii) a-ketoglutarate;(iv) mannonate; or(v) TAG54:4 / FA22:4,(b) the at least two biomarkers comprise:(i) 3-ureidopropionate and kynurenine;(ii) 3-ureidopropionate and a-ketoglutarate;(iii) 3-ureidopropionate and mannonate;(iv) 3-ureidopropionate and TAG54:4 / FA22:4;(v) kynurenine and a-ketoglutarate;(vi) kynurenine and mannonate;(vii) kynurenine and TAG54:4 / FA22:4;(viii) a-ketoglutarate and mannonate;(ix) a-ketoglutarate and TAG54:4 / FA22:4; or(x) mannonate and TAG54:4 / FA22:4,(c) the at least three biomarkers comprise:(i) 3-ureidopropionate, kynurenine, and a-ketoglutarate;(ii) 3-ureidopropionate, kynurenine, and mannonate;(iii) 3-ureidopropionate, kynurenine, and TAG54:4 / FA22:4;(iv) 3-ureidopropionate, a-ketoglutarate, and mannonate;(v) 3-ureidopropionate, a-ketoglutarate, and TAG54:4 / FA22:4;(vi) 3-ureidopropionate, mannonate, and TAG54:4 / FA22:4;(vii) kynurenine, a-ketoglutarate, and mannonate;(viii) kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4;(ix) kynurenine, mannonate, and TAG54:4 / FA22:4; or(x) a-ketoglutarate, mannonate, and TAG54:4 / FA22:4,(d) the at least four biomarkers comprise:(i) kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4;(ii) 3-ureidopropionate, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4;(iii) 3-ureidopropionate, kynurenine, mannonate, and TAG54:4 / FA22:4;(iv) 3-ureidopropionate, kynurenine, a-ketoglutarate, and TAG54:4 / FA22:4; or(v) 3-ureidopropionate, kynurenine, a-ketoglutarate, and mannonate, or(e) the at least five biomarkers comprise 3-ureidopropionate kynurenine, a-ketoglutarate, mannonate, and TAG54:4 / FA22:4.

156. The method of claim 155, wherein the at least one biomarker, at least two biomarkers, at least three biomarkers, or at least four biomarkers further comprises an additional biomarker selected from the group consisting of: X-26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2- methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

157. The method of claim 154, wherein:(a) the at least one biomarker comprises:(i) 3-ureidopropionate;(ii) kynurenine; or(iii) a-ketoglutarate;(b) the at least two biomarkers comprise:(i) 3-ureidopropionate and kynurenine;(ii) 3-ureidopropionate and a-ketoglutarate; or(iii) kynurenine and a-ketoglutarate; or(c) the at least three biomarkers comprise 3-ureidopropionate, kynurenine, and a-ketoglutarate.

158. The method of claim 157, wherein the at least one biomarker, at least two biomarkers, or at least three biomarkers further comprises an additional biomarker selected from the group consisting of: X- 26158, ALTUIL, 2'-deoxyuridine, X-12096, N-acetylmethionine, glycochenodeoxycholate 3-sulfate, tyrosine, 3-indoleglyoxylic acid, ribitol, aspartate transaminase, glucose, sphinganine, 2- methylcitrate / homocitrate, SAH, and 2'-O-methyluridine.

159. The method of any one of claims 1 -158, wherein the biological sample is a whole blood, serum, or plasma sample.

160. The method of claim 159, wherein the biological sample is a serum sample.161 . The method of any one of claims 1 -160 wherein the subject has a BMI less than 27.5 kg / m2.

162. The method of any one of claims 1 -160, wherein the subject has a BMI greater than or equal to 27.5 kg / m2.

163. A MASLD therapy for use in treatment of a subject having MASLD, the treatment comprising:(a) identifying the subject as having MASLD by determining that a level of at least one biomarker in a biological sample obtained from the subject is increased relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the MASLD therapy to the subject.

164. A MASLD therapy for use in monitoring efficacy of treatment in a subject having MASLD, the treatment comprising:(a) detecting an increase in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the MASLD therapy administered to the subject.

165. A MASLD therapy for use in monitoring efficacy of a treatment in a subject having MASLD, the treatment comprising:(a) detecting a decrease in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the MASLD therapy administered to the subject.

166. A MASLD therapy for use in a method of monitoring disease progression in a subject having MAFL, the method comprising:(a) detecting an increase in a level of at least one biomarker in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the increase in the level of the at least one biomarker indicates that MAFL has progressed to MASH, and wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'- O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4- hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 - phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2- FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the MASLD therapy to said subject.

167. A liver fibrosis therapy for use in treatment of a subject having liver fibrosis, the treatment comprising:(a) identifying the subject as having liver fibrosis by determining that a level of at least one biomarker in a biological sample obtained from the subject is increased relative to a reference level, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the liver fibrosis therapy to the subject.

168. A liver fibrosis therapy for use in monitoring efficacy of treatment in a subject having liver fibrosis, the treatment comprising:(a) detecting an increase in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the liver fibrosis therapy administered to the subject.

169. A liver fibrosis therapy for use in monitoring efficacy of treatment in a subject having liver fibrosis, the treatment comprising:(a) detecting a decrease in a level of at least one biomarker in a biological sample obtained from the subject relative to a reference level or a previous sample obtained from the subject, wherein the at least one biomarker is selected from the group consisting of: 1 -stearoyl-GPG (18:0), 2- methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3-indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2), DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3- sulfate, kynurenine, pantothenate, quinolinate, ribitol, S-adenosylhomocysteine (SAH), sphingadienine,sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4-FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4- FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) modifying the liver fibrosis therapy administered to the subject.

170. A liver fibrosis therapy for use in a method of monitoring disease progression in a subject having liver fibrosis, the method comprising:(a) detecting an increase in a level of at least one biomarker in a biological sample obtained from the subject relative to a previous sample obtained from the subject, wherein the increase in the level of the at least one biomarker indicates that the liver fibrosis has progressed to stage F2, F3, or F4 liver fibrosis, and wherein the at least one biomarker is selected from the group consisting of 1 -stearoyl-GPG (18:0), 2-methylcitrate / homocitrate, 2'-O-methyluridine, 3beta,7alpha-dihydroxy-5-cholestenoate, 3- indoleglyoxylic acid, 3-ureidopropionate, 4-hydroxychlorothalonil, a-ketoglutarate, alanine aminotransferase (ALT), aspartate transaminase (AST), cysteine s-sulfate, DAG(18:1 / 20:2),DAG(18:1 / 22:5), DCER(18:0), dimethylarginine (SDMA + ADMA), gamma-glutamyl transferase, glucose, glycochenodeoxycholate 3-sulfate, kynurenine, pantothenate, quinolinate, ribitol, S- adenosylhomocysteine (SAH), sphingadienine, sphinganine, sphingosine 1 -phosphate, Succinate, TAG44:1 -FA18:1 , TAG45:1 -FA15:0, TAG47:0-FA15:0, TAG48:0-FA14:0, TAG50:2-FA16:0, TAG50:4- FA18:2, TAG52:0-FA20:0, TAG54:4-FA18:1 , TAG54:4-FA22:4, TAG56:4-FA16:0, TAG58:7-FA22:5, TAG60:10-FA22:6, tyrosine, X-12096, and X-26158; and(b) administering the liver fibrosis therapy to said subject.171 . A computer implemented method for detection of a MASLD disease state in a subject comprising:(a) inputting a level of at least one biomarker selected from the group consisting of: 3- ureidopropionate, a-ketoglutarate, and kynurenine, determined in a biological sample obtained from a subject;(b) determining a risk probability score of the disease state in the subject by applying a predictive computer model previously trained using verified liver disease reference cases; and(c) automatically displaying the risk probability score on a graphical user interface.

172. A computer implemented method for detection of MASLD in a subject comprising:(a) inputting a the level of at least one biomarker selected from the group consisting of: 3- ureidopropionate, a-ketoglutarate, and kynurenine, determined in a biological sample obtained from a subject;(b) determining a risk probability score of MASLD in the subject by applying a predictive computer model previously trained using verified liver disease reference cases; and(c) automatically displaying the risk probability score on a graphical user interface.

173. A diagnostic device for determining the risk of MASLD disease progression in a subject comprising:(a) means to acquire data comprising a level of one or more biomarkers;(b) optionally, an analysis module operable to derive corrections of the data;(c) output means for producing a statistical model with the data including at least one metric of the level of the one or more biomarkers; and(d) output means indicating that if the metric indicates an improvement or resolution of one or more symptoms of MASLD, the treatment may be discontinued or administered at a reduced dose amount and / or frequency of administration.

174. A computerized system comprising a non-transient data storage circuit adapted to store measurements of a biomarker level produced by the method of any one of claims 1 -170.

175. Computer program product, comprising computer executable instructions for causing a computer, diagnostic arrangement, apparatus, or device to perform the steps of the method of any one of claims 1 - 170.

176. A computer-implemented method for classifying a human subject as having MASH F2-F3 comprising:(a) inputting a level of the at least one biomarker selected from the group consisting of: 3- ureidopropionate, a-ketoglutarate, and kynurenine, determined using a blood sample from a subject into a gradient-boosting machine previously trained using verified liver disease reference cases; and(b) outputting a binary classification that the subject does or does not exhibit MASH F2-F3 based on a comparison of the level of the at least one biomarker of the subject to the reference cases in the gradient boosting machine.

177. The computer implemented method of claim 176, wherein the method achieves an area-under-the- receiver-operating-characteristic curve (AUG) of at least 0.90.

178. A computer-implemented method for classifying a human subject as having MASH F4 comprising:(a) inputting a level of the at least one biomarker selected from the group consisting of: 3- ureidopropionate, a-ketoglutarate, and kynurenine, determined using a blood sample from a subject into a gradient-boosting machine previously trained using verified liver disease reference cases; and(b) outputting a binary classification that the subject does or does not exhibit MASH F4 based on a comparison of the level of the at least one biomarker of the subject to the reference cases in the gradient boosting machine.

179. The computer implemented method of claim 178, wherein the method achieves an AUC of at least 0.95.

180. The computer implemented method of any one of claims 176-179, wherein the determining of the level of at least one biomarker comprises liquid-chromatography tandem-mass-spectrometry (LC- MS / MS).181 . The computer implemented method of any one of claims 176-180, wherein the at least one biomarker is 3-ureidopropionate.

182. The computer implemented method of any one of claims 176-180, wherein the at least one biomarker is a-ketoglutarate.

183. The computer implemented method of any one of claims 176-180, wherein the at least one biomarker is kynurenine.

184. The computer implemented method of any one of claims 176-180, wherein the at least one biomarker comprises at least two biomarkers or at least three biomarkers.

185. The computer implemented method of claim 184, wherein:(a) the at least two biomarkers comprise:(I) 3-ureidopropionate and kynurenine;(ii) 3-ureidopropionate and a-ketoglutarate; or(Hi) kynurenine and a-ketoglutarate; or(b) the at least three biomarkers comprise 3-ureidopropionate, kynurenine, and a-ketoglutarate.

186. The computer implemented method of any one of claims 176-185, wherein the method further comprises inputting one or more pre-determined clinical variables into the gradient-boosting machine.

187. The computer implemented method of claim 186, wherein the one or more predetermined clinical variables comprise aspartate aminotransferase (AST) levels, alanine aminotransferase (ALT) levels, body mass index (BMI), the total number of components of the metabolic syndrome (MetS), platelet counts, albumin and / or age.

188. The computer implemented method of claim 176 or 177, wherein the method further comprises inputting one or more pre-determined clinical variables selected from AST levels, ALT levels, BMI, and the total number of components of the MetS.

189. The computer implemented method of claim 178 or 179, wherein the method further comprises inputting one or more pre-determined clinical variables selected from the group consisting of platelet count, albumin levels, BMI, and age.

190. A computer program product, comprising computer executable instructions for causing a computer, diagnostic arrangement, apparatus, or device to perform the steps of the computer implemented method of any one of claims 176-189.191 . A non-transitory computer-readable storage medium having stored thereon instructions which, when executed by one or more processors, cause the processor(s) to:(a) receive numerical inputs comprising:(i) at least one metabolite concentration selected from 3-ureidopropionate, a- ketoglutarate, and kynurenine, and optionally(II) a predetermined clinical variable selected from the group consisting of: AST levels, ALT levels, BMI, the total number of components of the MetS, platelet counts, albumin levels, and / or age;(b) standardize or normalize the inputs;(c) apply an initial CatBoost gradient-boosting decision-tree model having:(i) loss_function = “Logloss’’;(ii) iterations = 200;(iii) learning_rate = 0.05;(iv) depth = 6;(v) I2_leaf_reg = 1 , and / or(vi) bootstrap_type = “MVS", random_seed = 2, od_type = “Iter”, od_wait = 20;(d) generate a probability that the subject belongs to either fibrosis stage F2-F3 or stage F4 depending on a selected weight set and, optionally, comparing the probability to a threshold that maximizes Youden’s J statistic; and(e) output a binary classification of(i) “Eligible" or “Not-eligible” for MASH F2-F3, or(II) “Cirrhosis" or “No-cirrhosis” for MASH F4.

192. The non-transitory computer-readable storage medium of claim 191 wherein the program is implemented in the R programming language and delivered through a Shiny graphical user interface.

193. The non-transitory computer-readable storage medium of claim 191 or 192 wherein the CatBoost model weights are containerized within a Docker image digitally signed to ensure integrity.

194. The non-transitory computer-readable storage medium of any one of claims 191 -193 further comprising instructions that, prior to step (c), automatically perform hyper-parameter tuning by exhaustive search over iterations 200-600, learning_rate 0.05-0.20, depth 5-9, and I2_leaf_reg 1 -6, selecting the combination that maximizes area-under-the-receiver-operating-characteristic curve (AUC).

195. The non-transitory computer-readable storage medium of claim 194, wherein hyper-parameter tuning is terminated early if the validation AUC fails to improve for 20 consecutive iterations.

196. The non-transitory computer-readable storage medium of any one of claims 191 -195 wherein execution time for steps (b)-(f) is less than one second on a system having four virtual CPU cores.

197. A computer-implemented method comprising:(a) deploying the non-transitory computer-readable storage medium of any of claims 191 -196 within a hospital information-technology environment;(b) receiving subject data via an electronic health record interface and simplifying its structure into a comma-separated value (CSV) file;(c) executing the Catboost model; and(d) writing a binary result back to the electronic medical record without human intervention.

198. The computer-implemented method of claim 197 wherein the model weights correspond to the MASH F2-F3 classifier of claim 176 and achieve an AUC of at least 0.90 on unseen data.

199. The computer-implemented method of claim 197 wherein the model weights correspond to the MASH F4 classifier of claim 178 and achieve an AUC of at least 0.95 on unseen data.

200. A computer-implemented method for diagnosing MASH F2-F3 in a human subject, comprising:(a) obtaining from the subject a numeric value for each of the following seven biomarkers:(i) ALT;(ii) TotalMetS;(iii) AST;(iv) BMI;(v) 3-ureidopropionate;(vi) a-ketoglutarate; and(vii) kynurenine;(b) assigning the value for each biomarker to one of three risk tiers (low, grey, or high) by comparing the value to two pre-determined numeric limits, LIM1 and LIM2, wherein LIM1 < LIM2 for the biomarker;(c) computing a cumulative fibrosis score (S) by adding 0 points for a low tier assignment, 1 point for a grey tier assignment and 2 points for a high tier assignment for each biomarker; and(d) classifying the subject as having MASH F2-F3 when S > 6, and as stage F0-F1 otherwise.201 . The method of claim 200, wherein the LIM1 and LIM2 limits for each biomarker are:

202. The method of claim 201 , wherein a missing or non-detectable value for ALT, AST or BMI is automatically treated as low risk for this specific variable.

203. The method of any one of claims 200-202, wherein the classification in step (d) further outputs the probability (P) of MASH F2-F3 (P(F2-F3)) produced by a gradient-boosted ensemble of depth-8 symmetric decision trees trained on a reference cohort, and displays both S and P(F2-F3) to the clinician.

204. The method of claim 203, wherein the ensemble comprises 200 trees and is trained with the CatBoost algorithm.

205. The method of claim 200, wherein the limits LIM1 and LIM2 are the two most frequently used split thresholds mined from the trained ensemble.

206. The method of claim 200, wherein S > 8 triggers automatic referral for liver biopsy or specialist evaluation.

207. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, perform the method of any one of claims 200-206.

208. A diagnostic system comprising:(a) an input interface configured to receive the biomarker values;(b) a processor configured to execute the instructions of claim 207; and(c) a user interface configured to display the risk-tier assignments, cumulative score S, and P(F2- F3).

209. The method of claim 200, wherein 3-ureidopropionate is used as a surrogate marker of abnormal uracil catabolism, kynurenine as a surrogate of host-microbiome tryptophan metabolism, and a- ketoglutarate as a surrogate of tricarboxylic acid cycle dysregulation, thereby covering any biochemical analyte or analytical platform that captures the same metabolic pathways.

210. A method according to any one of claims 200-209, further comprising repeating steps (a)-(d) after initiation of an anti-MASH therapy and flagging a responder status when the follow-up cumulative score S2decreases by >2 points relative to baseline Si.211 . The method of claim 200, wherein steps (a)-(d) are applied to predict a binary clinical outcome selected from: cirrhosis (MASH F4), cardiovascular event, hepatocellular carcinoma, liver transplantation, or all-cause mortality.

212. The method of claim 200, wherein steps (a)-(d) are applied to any disease state for which a labelled training set exists, the two-limit per variable abstraction being automatically learned from the underlying decision-tree ensemble.

213. The method of any of claims 200-212, further comprising providing, on a non-transitory computer- readable medium, executable source code that:(a) loads a binary CatBoost-format model file (.cbm);(b) parses the oblivious_trees object to extract all split borders for each float feature;(c) identifies, for every feature, the two borders that occur with highest frequency;(d) stores said borders as limits LIM1 and LIM2 in a human-readable table; and(e) implements steps (b)-(d) of claim 200 using those limits.

214. The computer-readable medium of claim 213, wherein the source code is written in a language selected from Python, R, JavaScript, or any combination thereof, and the CatBoost model is loaded via a call to the CatBoost().load_model() API.

215. The method of claim 213 or 214, wherein the source code automatically generates a self-contained Python or R script that hard-codes the LIM1 and LIM2 limits and the scoring logic, thereby enabling deployment of a lightweight interpretability layer without requiring the full CatBoost runtime.

216. The method of any one of claims 200-215, further comprising a training pipeline which (a) ingests a labelled dataset comprising biomarker vectors and binary clinical outcomes, (b) trains a gradient-boosted ensemble of symmetric decision trees with depth < 8, and (c) serializes the trained ensemble into said binary CatBoost-format file.

217. The method of claim 216, wherein steps (a)-(c) are executed inside an automated continuous- integration workflow that triggers re-training and re-serialization whenever new outcome-labelled data are appended to the training set.

218. A kit comprising:(a) analytical reagents or calibrated sensors for quantifying at least one biomarker selected from ALT, AST, a-ketoglutarate, 3-ureidopropionate, TotalMetS components, BMI, and kynurenine; and(b) the computer-readable medium of any of claims 213-215, wherein the kit, when used according to the instructions, produces the cumulative fibrosis score S of claim 200.

219. The method of any of claims 200-218, further comprising presenting to a clinician a graphical user interface that simultaneously displays (a) the per-biomarker tier assignments, (b) the cumulative score S,(c) the model-derived P(F2-F3), and (d) a recommended clinical action selected from lifestyle intervention, pharmacologic therapy, further imaging, or liver biopsy.

220. The method of any of claims 200-219, wherein the same source-code framework is applied to a training dataset for a second binary outcome selected from cardiovascular event, hepatocellular carcinoma, liver-related mortality, overall mortality, response vs non-response to anti-MASH therapy, or development of type 2 diabetes, and wherein the resulting two-limit per-variable scoring table is generated and deployed via the process of claim 213.221 . A computer-implemented method for identifying cirrhotic fibrosis stage F4 in a human subject, comprising:(a) obtaining numeric values for each of the following seven biomarkers:(I) platelet count;(ii) BMI;(iii) serum albumin;(iv) age;(v) 3-ureidopropionate;(vi) a-ketoglutarate; and(vii) kynurenine;(b) assigning each biomarker to one of three risk tiers — low, grey or high — by comparing the value with two limits LIM1 < LIM2 specific to that biomarker;(c) computing a cumulative cirrhosis score C by adding 0 points for a low-tier assignment, 1 point for a grey-tier assignment and 2 points for a high-tier assignment for every biomarker; and(d) classifying the subject as cirrhotic (F4) when C > 7, and as non-cirrhotic (F0-F3) otherwise.

222. The method of claim 221 , wherein the tier limits are:

223. The method of claim 221 or 222, wherein the limits LIM1 and LIM2 are the two most frequently used split borders extracted from a gradient-boosted ensemble of depth-8 symmetric decision trees.

224. The method of claim 223, wherein the ensemble comprises 200 trees and is stored in CatBoost binary format.

225. The method of any of claims 221 -224, wherein missing platelet, BMI or albumin values are automatically treated as low risk.

226. The method of claim 221 , wherein C > 9 triggers automatic referral for portal-hypertension work-up or transplant evaluation.

227. A non-transitory computer-readable medium comprising source code that:(a) loads the CatBoost model file of claim 224;(b) parses all float-feature splits to identify the two highest-frequency borders per feature;(c) writes those borders into a human-readable table as in claim 222; and(d) executes steps (b)-(d) of claim 221 on patient data.

228. The medium of claim 227, wherein the code automatically regenerates the table and redeploys the scoring script whenever the model is retrained on an updated dataset.

229. The method of any of claims 221 -228, wherein the cirrhosis-score framework is further applied to predict a binary outcome selected from decompensation, variceal bleeding, hepatocellular carcinoma, liver-transplant-free survival, or all-cause mortality (using the same two-limit abstraction learned from any CatBoost ensemble trained on corresponding labelled datasets).

230. The method of any of claims 200-229, wherein the biomarker-classification and scoring steps are implemented in an interpreted-language script that:(a) imports the catboost, dplyr and pROC libraries;(b) extracts a training frame and a validation frame each containing at least one metabolomic feature and the binary outcome;(c) performs hyper-parameter grid search over iterations, learning_rate, depth and I2_leaf_reg;(d) selects the parameter set that maximizes validation AUC; and(e) trains a final CatBoost model with that parameter set.231 . The method of claim 230, wherein the grid comprises 200-600 iterations, learning-rates 0.05-0.20, tree depth 5-9 and L2-leaf-regularisation 1-6.

232. The method of claim 230 or 231 , further comprising:(a) evaluating the trained model on a combined dataset to obtain per-subject probabilities;(b) selecting an operating threshold that maximizes Youden’s J statistic on the ROC curve; and(c) calculating accuracy, sensitivity, specificity, positive-predictive value and negative-predictive value at that threshold.

233. The method of any of claims 230-232, wherein the script serializes the trained model to a CatBoost binary file named by concatenating the outcome label and selected feature names with underscore separators and the extension “.cbm.”234. The method of claim 233, further comprising storing a CSV file of the selected hyper-parameters and a plain-text printout of the rounded performance metrics.

235. The method of any of claims 230-234, wherein the same script is re-executed automatically whenever the training dataset is modified, thereby regenerating the model file, parameter CSV and metrics without manual intervention.

236. The method of any of claims 230-235, wherein the grid-search, model-training, threshold-selection and metric-reporting steps are parallelized across multiple CPU cores via a doParallel or foreach backend.

237. A diagnostic kit comprising (i) reagents or sensors for quantifying one or more biomarkers in any of claims 200-229 and (ii) the computer-readable medium of any of claims 230-234, whereby the kit produces both raw probability output and thresholded class assignment in real time.

238. The method of any of claims 200-237, further comprising a feature-selection workflow that:(a) performs recursive-feature-elimination (RFE) with a CatBoost estimator to obtain per-feature importance;(b) performs principal-component analysis (PCA) on the same numeric predictors and records the Euclidean loading magnitude of each feature on the first two principal components; and(c) fits an elastic-net logistic-regression model and records the absolute value of each non-zero coefficient.

239. The method of claim 238, wherein the RFE step uses ten-fold cross-validation and a CatBoost base learner trained for 100 iterations with a log-loss objective.

240. The method of claim 238 or 239, wherein the PCA step calculates the loading magnitude as ^(PC12+ PC22) for every feature.241 . The method of any of claims 238-240, wherein the elastic-net step selects the mixing parameter a via grid search from 0.1 to 0.9 and the regularization strength A via internal cross-validation.

242. The method of any of claims 238-241 , further comprising normalizing each of the three importance measures to its own maximum, summing the normalized values to obtain a combined score, and ranking features by that combined score.

243. The method of claim 242, wherein the top-N ranked features (N < 20) are merged with a predetermined list of clinical covariates to form the final predictor set for model training.

244. The method of any of claims 238-243, wherein the ranked feature list and the combined scores are exported to a CSV file titled “top_selected_features_per_combined_score.csv”.

245. The computer-readable medium of any of claims 233, 234, and 237, further comprising storing an R script that embodies the workflow of claims 238-244 and serializes the resulting CatBoost model to a filename derived from the outcome label concatenated with “FS.cbm”.

246. A kit comprising:(a) LC-MS / MS calibration standards for 3-ureidopropionate, kynurenine, and a-ketoglutarate;(b) reagents for protein precipitation, drying, and re-constitution; and(c) non-transitory computer-readable media storing the gradient-boosting weight sets of claims 191 -245.

Images