Methods of assessment of pre-clinical alzheimer's disease

Abstract

Described herein are methods of assessing the risk of dementia in a subject using an odor-based test, or "smell test". These methods can be self-administered by the subject under the instruction of a computing device via a user interface. Methods of administering a smell test as described herein can optionally be combined with measuring and quantifying the level of at least one biomarker for dementia (e.g., CXCL10, CCL2, IL-6) in a biological sample collected from the subject. Also provided are methods to treat dementia (e.g., dementia associated with Alzheimer's disease, dementia associated with a subset of Alzheimer's disease patients with TDP-43 pathology).

Description

The present disclosure provides methods for administration of a smell test. Methods of administering the smell test can utilize at least one odor card as disclosed herein, at least one test guide as disclosed herein, and / or a computing device as disclosed herein. A smell test as understood herein can encompass at least one odor sub-test. In some embodiments, an odor sub-test comprises at least one odor intensity test(s) (OIT), at least one odor percept identification test (OPID, or odor identification test), at least one percepts of odor episodic memory test (POEM, or odor episodic memory test), at least one odor discrimination test (OD), and / or any combination thereof. Combinations can include variables such as a subject incorrectly responding to one or more (e.g., two or more, etc.) of the odor identification / discrimination tests. Other combinations can include engineered variables such as a subject incorrectly responding to one or more (e.g., two or more, etc.) of the odor identification / discrimination tests and providing a low score (e.g., relative to a predetermined threshold) to one or more odor intensity tests. It is to be understood by one of ordinary skill in the art that various other combinations can also be included to develop engineered variables, including combinations of one or more responses from the symptom questionnaire. In some embodiments, a subject can self-administer a smell test. In some embodiments, a subject can self-administer a smell test under the instruction provided by the user interface generated by a computing device as disclosed herein. In some embodiments, when conducting the one or more smell tests according to the steps described herein, the smell test responses or results from the subject are compiled in a database for storage, further processing, or calculations according to the software modules as disclosed herein. Odor Identification Test An odor identification test as disclosed herein can include the tests illustrated in the top and middle panels of FIG. IB and can utilize odor cards A, B and / or C as shown in FIG. 1A. The odor identification test employs a plurality of odors (e.g., nine and / or eighteen odors) found to be predictive for conversion from Mild Cognitive Impairment to a neurodegenerative disease, such as Alzheimer's disease. This test is referred to as an odor percept identification test (e.g., OPID9 or OPID18) because it requires the subject to identify an odor based on the subject's odor percept, or what the subject remembers, and by so doing, builds the subject's working memory into that test. In an odor identification test disclosed herein, the subject can be instructed (e.g., by a computing device via a user interface) to prepare to sniff at least one odor from the odor card. In some embodiments, after the odor presentation, a subject is instructed to rate the intensity of the odor on a Likert scale from 0 to 10. An example of such instruction is demonstrated in FIG. 4A. In an odor identification test disclosed herein, after rating the intensity of the odor, the subject can be presented (e.g., by a computing device via a user interface) with one or more odor names and instructed and to choose the label that best represented the odor they smelled. In some embodiments, a subject is then instructed to rate their confidence in their identification of the odor from the one or more odor names provided by the user interface. For example, a subject is instructed to rate their confidence in their identification of the odor using the following scale: “I Guessed,” “I Narrowed Down to Three,” “I Narrowed Down to Two,” or “I Am Certain.” An example of such instructions is demonstrated in FIG. 4B. Responses provided by the subject can be analyzed and scored, forwarded to the software modules / algorithms disclosed herein for calculating and displaying the results of the test, and / or may be stored within the database. In some embodiments, responses provided by the subject can be used to calculate an average intensity score, wherein the average intensity score can be derived from the mean intensity ratings of the plurality of odors on the Likert scale. In some embodiments, responses provided by the subject can be used to calculate a score for the odor identification test, wherein the score can be calculated as the total number of correctly identified odors. In some embodiments, responses provided by the subject can be used to generate a confidence metric for the smell test. A confidence metric for the smell test is calculated by the computing device as the total number of odors identified correctly by the subject where the subject did not evaluate their confidence in their odor identification decision by selecting “I Guessed” when prompted by the computing device via the user interface. Once calculated, a confidence metric for the smell test can be compared to a predetermined threshold. A predetermined threshold as used herein can be determined by a machine learning algorithm (e.g., by a computing device) based on a plurality of previously entered smell test performance data from healthy subjects (i.e., subjects with no cognitive impairment). A smell test for use herein can comprise more than one odor identification tests. In some embodiments, a smell test disclosed herein can comprise two odor identification tests. In some embodiments, a smell test disclosed herein can comprise two odor identification tests wherein the odor identification tests are performed subsequently. In some embodiments, a smell test disclosed herein can comprise two odor identification tests wherein the odor identification tests are performed subsequently with at least a 10-minute break between administering each test. In some embodiments, a smell test disclosed herein can comprise two odor identification tests wherein a first odor identification test is performed, followed by a different odor sub-test (e.g., an odor episodic memory test), and then a second odor identification test. In some embodiments, a smell test disclosed herein can comprise two odor identification tests wherein a first odor identification test is comprised of odors that are not presented in the second odor identification test. In some embodiments, a smell test disclosed herein can comprise two odor identification tests wherein a first odor identification test is comprised of a different plurality of odors that are more / less than the number of odors provided in the second odor identification test. In some embodiments, an odor identification test is comprised of at least nine odors (OPID9). In some embodiments, an odor identification test is comprised of nine odors (OPID9) consisting of menthol, clove, leather, strawberry, lilac, pineapple, smoke, soap, and grape. In some embodiments, an odor identification test is comprised of at least eighteen odors (OPID18). In some embodiments, an odor identification test is comprised of eighteen odors (OPID18) consisting of menthol, clove, leather, strawberry, lilac, pineapple, smoke, soap, grape, coffee, peach, chocolate, orange, dirt, banana, lemon, bubble gum, and rose. In some embodiments, an OPID9 test is performed before an OPID18 test. Odor Episodic Memory Test An odor episodic memory test (e.g., POEM) as disclosed herein can include the tests illustrated in the top and middle panels of FIG. IB and can utilize odor cards A, B and / or C as shown in FIG. 1A. An odor episodic memory test disclosed herein, can comprise the subject being instructed (e.g., by a computing device via a user interface) to prepare to sniff at least one odor from the odor card. The method of administering an odor episodic memory test is similar to the original odor identification test described above except for asking the subject the question (e.g., provided by the use interface) following the presentation of the odor: “Did you smell this odor in the previous test?” An example of such instructions is demonstrated in FIG. 4C. An odor episodic memory test disclosed herein measures a subject's ability to remember previously presented odors after a delay of a fixed period (e.g., 10 minutes) with no clues, and further measures how accurate the subject was in identifying the odor. Put another way, subjects can be asked if the odor presented was new or had been presented in the earlier odor identification test. In some configurations, the instruction phase of the test may include an explicit indication that the new / old designation refers to the current testing session and not to a broader lifetime exposure. In some embodiments, following this measure of episodic or autobiographical odor memory, a subject subjects can be presented with a plurality of odor names, and asked to choose which name associated with their memory of the odor percept they experienced at the start of the smell test (e.g., from an odor identification test administered before the odor episodic memory test). The odor memory / identification response (e.g., yes or no) can be received from the subject, optionally or in addition to, the odor name associated with their memory / identification of the odor percept experienced. These responses may be analyzed and scored, forwarded to the software modules / algorithms disclosed herein for calculating and displaying the results of the test, and / or can be stored within a database as disclosed herein. In some embodiments, subject response can be used to calculate a POEM index. In some embodiments, a POEM index can be calculated as the difference between the proportion of correct and incorrect recognitions, with scores ranging from -1 to 1. Odor Discrimination Test An odor discrimination test (e.g., OD) as disclosed herein can include the tests illustrated in the bottom panel of FIG. IB and can utilize odor cards D and / or E as shown in FIG. 1A. An odor discrimination test disclosed herein, can comprise the subject being instructed (e.g., by a computing device via a user interface) to prepare to sniff at two odors from the odor card consecutively. In some embodiments, the subject is instructed to smell an odor on the odor card for a fixed period, using techniques analogous to those previously disclosed above. At the end of the specified time period (e.g., two seconds), the subject is instructed to smell an odor on the odor card for the same fixed period, one after the other, using techniques analogous to those previously disclosed. In some embodiments, a subject can then be asked if the two odors presented were the same or different (yes / no). An example of such instructions is demonstrated in FIG. 4D. These responses may be analyzed and scored, forwarded to the software modules / algorithms disclosed herein for calculating and displaying the results of the test, and / or can be stored within a database as disclosed herein. In some embodiments, subject response can be used to calculate an odor discrimination (OD) score. In some embodiments, an OD score can be calculated as the total number of correctly discriminated odor pairs. Kits The present disclosure provides kits for use in performing any of the methods for administering a smell test as disclosed herein. In some embodiments, a kit can comprise at least one odor card and at least one test guide. In some embodiments, a kit can further include an instructional card and / or a card with information to direct a subject to a website comprising instructions. In some embodiments, a kit can further include one or more materials needed for collecting, transporting, and preserving a biological sample. Methods of Use Provided herein are methods of assessing risk of dementia in a subject. Also provided herein are methods of assessing a subject for their potential responsiveness to a therapy that comprises administration of a TYK2 inhibitor. In some embodiments, a subject identified as at risk of dementia according to the methods disclosed herein can be administered at least one therapy for treating, preventing (reducing further risk of), and / or ameliorating a neurodegenerative disease, e.g., Alzheimer’s disease. Subject Populations Methods of assessing risk of dementia and / or responsiveness to a TYK2 inhibitor, as provided in the present disclosure, are suitable for use in a subject in need thereof. “Subject” and “patient” refer to either a human or non-human, such as mammals, e.g., vertebrates, e.g., primates. In some embodiments, the subject is a human or a non-human veterinary subject such as a non-human primate, cat, dog, horse, cow, goat, or rabbit. In some embodiments, the subject is an adult human. In some embodiments, the subject is an adult human that is at least about or about 55 years of age. In some embodiments, the subject is an adult human that is at least about or about 55-100, 55-95, or 55-90 years of age. In some embodiments, a subject in need thereof is a subject suspected of having or has been diagnosed as having cognitive impairment. In some embodiments, a subject in need thereof has, is suspected of having, or is at risk of having a neurodegenerative disease associated with a TDP-43 pathology. Methods of diagnosing TDP-43 pathology in a subject are known in the art (see, e.g., Lopez-Carbonero et al., TranslNeurodegener. 2024 Jun 3; 13(1):29; Ducharme et al., Am J Geriatr Psychiatry. 2024 Jan;32(l):98-113) and are suitable for use herein. In some embodiments, a subject in need thereof has, is suspected of having, or is at risk of having Alzheimer’s disease. Methods of diagnosing AD in a subject are known in the art (see, e.g., Podhorna et al., Alzheimers Res Ther. 2016 Feb 12;8:8; van Oostveen et al., IntJMol Sci (2021) Feb 20;22(4):2110; Dinis-Oliveira et al., Forensic Sci Res (2017) Jan 16; 1 (1):42-51; Teunissen et al., Alzheimers Dement (2025) Jan;21(l):el4397) and are suitable for use herein. In some embodiments, a subject in need thereof has a loss-of-function SNP in TYK2 (rs34536443). In some embodiments, a subject in need thereof has as subset of Alzheimer’s disease associated with a TDP-43. Assessment of Risk of Dementia Methods provided herein comprise administering a smell test for assessment of risk of dementia in a subject. Dementia can result from neurodegeneration with Alzheimer's Disease (AD) being the most common cause of dementia. As such, methods provided herein also comprise administering a smell test for assessment of risk of AD in a subject. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise administering at least one odor identification test, at least one odor episodic memory test, and / or at least one odor discrimination test. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise administering at least one odor identification test. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise administering at least two odor identification tests. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise administering at least two odor identification tests wherein at least one odor episodic memory test is administered between the administration of the at least two odor identification tests. A method of administering a smell test for assessment of risk of dementia (e.g., of risk of dementia associated with AD) in a subject can comprise administering at least one odor identification test. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise instructing a subject to utilize an odor card disclosed herein, instructing a subject to identify an odor delivered by the odor card, and asking the subject to evaluate their confidence in their odor identification decision. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise instructing a subject to utilize an odor card disclosed herein, instructing a subject to identify an odor delivered by the odor card, and asking the subject to evaluate their confidence in their odor identification decision, wherein the administering is by the computing device via the user interface. In some embodiments, a subject is instructed to repeat the method above for each of the odors provided on the odor card. In some embodiments, a subject performs the method above for each of nine odors provided on the odor card. In some embodiments, a subject performs the method above for each of eighteen odors provided on the odor card. A method of administering a smell test for assessment of risk of dementia (e.g., of risk of dementia associated with AD) in a subject can further comprise administering at least one odor episodic memory test. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise instructing a subject to utilize an odor card disclosed herein, asking the subject if the odor delivered in the odor card was delivered in the earlier used odor card / earlier odor identification test, instructing a subject to identify an odor delivered by the odor card, and asking the subject to evaluate their confidence in their odor identification decision. In some embodiments, a method of administering a smell test for assessment of risk of dementia in a subject can comprise instructing a subject to utilize an odor card disclosed herein, asking the subject if the odor delivered in the odor card was delivered in the earlier used odor card / earlier odor identification test, instructing a subject to identify an odor delivered by the odor card, and asking the subject to evaluate their confidence in their odor identification decision, wherein the administering is by the computing device via the user interface. A subject’s confidence in their odor identification decision can be used to calculate, by the computing device, a confidence metric for the smell test. Using any of the software modules / algorithms as disclosed herein, a confidence metric for the smell test can be calculated by the computing device as the total number of odors identified correctly by the subject where the subject did not evaluate their confidence in their odor identification decision by selecting “I Guessed” when prompted by the computing device via the user interface. In some embodiments, the calculated confidence metric for the smell test can be compared to a predetermined threshold. In some embodiments, the calculated confidence metric for the smell test can be compared to a predetermined threshold by a computing device. In some embodiments, a predetermined threshold is determined by a machine learning algorithm based on a plurality of previously entered smell test performance data from healthy subjects (i.e., cognitively normal subjects). After administering a smell test according to the methods disclosed herein, a subject is identified as high risk for dementia when the calculated confidence metric for the smell test is lower than the predetermined threshold. In some embodiments, after administering a smell test according to the methods disclosed herein, a subject is identified as high risk for AD and / or AD-associated dementia when the calculated confidence metric for the smell test is lower than the predetermined threshold. Methods of administering a smell test as disclosed herein can assess neuroanatomical volume in at least one brain region in a subject. Neuroanatomical volume refers to the spatial measurement of different brain structures and regions, typically in cubic centimeters (cm3) or milliliters (mL). Neuroanatomical volume calculations from MRI scans, as described herein, can involve software-based automated methods, manual tracing, or a combination of both (see, e.g., Giorgio & Stefano, Journal of Magnetic Resonance Imaging. 2013;37:1-14). Determination of neuroanatomical volumes can be used as a quantitative measure of AD severity and to diagnose subjective cognitive decline (SCD), mild cognitive impairment (MCI), and AD dementia in a subject. In some embodiments, a smell test can be administered to identify a decline in neuroanatomical volume in at least one brain region in a subject. In some embodiments, after administering a smell test according to the methods disclosed herein, a subject is identified as having a decline in neuroanatomical volume in at least one brain region when the calculated confidence metric for the smell test is lower than the predetermined threshold. In some embodiments, after administering a smell test according to the methods disclosed herein, a subject is identified as having a decline in neuroanatomical volume in at least one brain region when the calculated score for the smell test is lower than the predetermined threshold. In some embodiments, the predetermined threshold is determined from smell test performance data collected from a plurality of healthy subjects (i.e., cognitively normal subjects). In some embodiments, the healthy subjects (i.e., cognitively normal subjects) have normal computed brain volumes as measured from MRI scans. In some embodiments, an OPID9 smell test as disclosed herein can be administered to identify a decline in neuroanatomical volume in at least one brain region in a subject. In some embodiments, an OPID18 smell test as disclosed herein can be administered to identify a decline in neuroanatomical volume in at least one brain region in a subject. In some embodiments, the at least one brain region in the subject can be the hippocampus, the amygdala, and / or both. In some embodiments, the at least one brain region in the subject can be the left hippocampus, the right hippocampus, the left amygdala, the right amygdala, or any combination thereof. In some embodiments, the identification of a decline in neuroanatomical volume in at least one brain region in the subject correlates to cognitive decline in the subject. In some embodiments, the identification of a decline in neuroanatomical volume in at least one brain region in the subject correlates to dementia in the subject. In some embodiments, the identification of a decline in neuroanatomical volume in at least one brain region in the subject correlates to AD in the subject. In some embodiments, a subject is diagnosed with subjective cognitive decline (SCD) following administration of the smell test, wherein the smell test identified a decline in neuroanatomical volume in at least one brain region (e.g., the hippocampus, the amygdala). In some embodiments, a subject is diagnosed with mild cognitive impairment (MCI) following administration of the smell test, wherein the smell test identified a decline in neuroanatomical volume in at least one brain region (e.g., the hippocampus, the amygdala). In some embodiments, a subject is diagnosed with dementia / AD dementia following administration of the smell test, wherein the smell test identified a decline in neuroanatomical volume in at least one brain region (e.g., the hippocampus, the amygdala). Methods of administering a smell test as disclosed herein can be combined with one or more other methods for assessing risk of dementia / risk of AD. In some embodiments, a smell test can be administered in addition to an assessment of a subject’s cognitive impairment as measured by ADAS-cog or a variant thereof (e.g., ADAS-Cog 3, ADAS-Cog 5, the original ADAS-Cog 11, and ADAS-Cog 13), Mini Mental State Examination (MMSE), or a combination thereof (see, e.g., Podhorna et al., Alzheimers Res Ther (2016) Feb 12;8:8). In some embodiments, a smell test can be administered in addition to one or more suitable imaging techniques (e.g., computed topography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET)) (see, e.g., van Oostveen et al., Ini J Mol Sci (2021) Feb 20;22(4):2110). In some embodiments, a smell test can be administered in addition to measurement of one or more biomarkers. Biomarkers can include biomolecules, for example but not limited to, carbohydrates, proteins, lipids to genes, DNA, RNA, platelets, enzymes, hormones, and the like. A biomarker for use herein can be measured in a biological sample collected from a subject according to methods known in the art (see, e.g., Dinis-Oliveira et al., Forensic Sci Res (2017) Jan 16; 1(1):42-51). In some embodiments, a biological sample for use herein can comprise whole blood, serum, plasma, and / or cerebrospinal fluid (CSF). In some embodiments, a smell test can be administered in addition to measurement of one or more biomarkers as identified in the present disclosure. In some embodiments, the one or more biomarkers as identified in the present disclosure can comprise C-X-C motif chemokine ligand 10 (CXCL10), C-C motif chemokine ligand 2 (CCL2), and interleukin 6 (IL-6). Methods of assessing these biological samples can comprise (1) collecting a biological sample (e.g., whole blood, serum, or plasma) from a subject, (2) quantifying the level of CXCL10, CCL2, and / or IL-6 in the biological sample; and (3) comparing the level of CXCL10, CCL2, and / or IL-6 to a predetermined threshold, wherein the predetermined threshold is determined from levels of the corresponding biomarker (e.g., of CXCL10, CCL2, and / or IL-6) as assessed from biological samples collected from a plurality of healthy subjects. In some embodiments, the level of CXCL10, CCL2, and / or IL-6 can be measured by subjected the biological sample to and immunoassay (e.g., sandwich immunoassay (V-PLEX), ELISA, flow cytometry). In some embodiments, comparing the level of CXCL10, CCL2, and / or IL-6 to a predetermined threshold can be performed by a computing device as disclosed herein, wherein a user inputs the measured levels of CXCL10, CCL2, and / or IL-6 into the database. In some embodiments, a subject can be identified as at high risk for dementia when the level of least one biomarker for dementia (e.g., CXCL10, CCL2, and / or IL-6) is higher than the predetermined threshold. In some embodiments, a subject can be identified as at high risk for dementia when the level of CXCL10 is higher than the predetermined threshold for CXCL10. In some embodiments, a subject can be identified as at high risk for dementia when the level of CCL2 is higher than the predetermined threshold for CCL2. In some embodiments, a subject can be identified as at high risk for dementia when the level of IL-6 is higher than the predetermined threshold for IL-6. In some embodiments, a subject can be identified as at high risk for dementia when the levels of CXCL10 and CCL2 are higher than the predetermined thresholds for CXCL10 and CCL2, respectively. In some embodiments, a subject can be identified as at high risk for dementia when the levels of CXCL10 and IL-6 are higher than the predetermined thresholds for CXCL10 and IL-6, respectively. In some embodiments, a subject can be identified as at high risk for dementia when the levels of CCL2 and IL-6 are higher than the predetermined thresholds for CCL2 and IL-6, respectively. In some embodiments, a subject can be identified as at high risk for dementia when the levels of CXCL10, CCL2, and IL-6 are higher than the predetermined thresholds for CXCL10, CCL2 and IL-6, respectively. Assessment of Responsiveness to TYK2 Inhibitors Methods provided herein comprise administering a smell test for assessing if a subject in need thereof will be responsive to treatment with at least one TYK2 inhibitor. Neuroinflammation is a pathological feature of many neurodegenerative diseases, including Alzheimer’s disease (AD)1,2 and amyotrophic lateral sclerosis (ALS)3, raising the possibility of common therapeutic targets. Cytoplasmic doublestranded RNA (cdsRNA) is spatially coincident with cytoplasmic pTDP-43 inclusions in neurons of patients with C9ORF72-mediated ALS4. CdsRNA triggers a type-I interferon (IFN-I)-based innate immune response in human neural cells, resulting in their death4. As provided in the present disclosure, cdsRNA is present and spatially coincident with pTDP-43 cytoplasmic inclusions in brain cells of patients with AD pathology, and that IFN-I response genes are significantly upregulated in brain regions affected by AD. cdsRNA also accumulates in two human cellular models of TDP-43 pathology. Cryptic exon (CE) detection as a proxy of pTDP-43 inclusions was used herein and demonstrated, using a machine-learning pipeline DRIAD-SP (Drug Repurposing In Alzheimer’s Disease with Systems Pharmacology), that the FDA-approved JAK inhibitors baricitinib and ruxolitinib that block interferon signaling show a protective signal only in a subset of brains with elevated CE expression. Furthermore, a CRISPR screen of cdsRNA-mediated death in differentiated human neural cells revealed the JAK family member TYK2 as a top hit. The selective TYK2 inhibitor deucravacitinib, an FDA-approved drug for psoriasis, rescued toxicity elicited by cdsRNA. Converging evidence from cell-based assays and from human observational studies of a TYK2-activity-reducing single nucleotide polymorphism (SNP) support CXCL10 as a candidate predictive biomarker for cdsRNA-related neurodegenerative diseases. Together, the present disclsosure provides parallel neuroinflammatory mechanisms between dementia patients with TDP-43-associated AD pathology and ALS that depend on TYK2, a possible diseasemodifying target of these thus far incurable neurodegenerative diseases. The present disclosure provides (1) inflammatory biomarkers for identification of neurodegenerative diseases with cdsRNA-pathology and (2) a correlation of the smell test with dementia and TDP-43 pathology - both of which can aid in identifying patients that could benefit from therapies mediating TYK2 activity (e.g., TYK2 inhibitor). As shown herein, cdsRNA-induced innate immune responses contribute to disease progression in the subset of AD patients with co-existing TDP-43 pathology and also in other neurodegenerative diseases, including ALS and FTD. Methods of detection of the activation of cdsRNA-mediated neurodegeneration in the patients affords the possibility of precision medicine in clinical trials for AD and other neurodegenerative diseases. Provided herein are methods for selecting a subject for treatment with a TYK2 inhibitor. In some embodiments, the method comprises administering a smell test according to the methods disclosed herein. In some embodiments, the method comprises selecting the subject for treatment with a TYK2 inhibitor if the calculated confidence metric for the smell test is lower than the predetermined threshold. In some embodiments, a method for selecting a subject for treatment with a TYK2 inhibitor method comprises quantifying a level of at least one biomarker for dementia in a biological sample collected from the subject according to the methods disclosed herein. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the level of least one biomarker for dementia (e.g., CXCL10, CCL2, and / or IL-6) is higher than the predetermined threshold. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the level of CXCL10 is higher than the predetermined threshold for CXCL10. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the level of CCL2 is higher than the predetermined threshold for CCL2. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the level of IL-6 is higher than the predetermined threshold for IL-6. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the levels of CXCL10 and CCL2 are higher than the predetermined thresholds for CXCL10 and CCL2, respectively. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the levels of CXCL10 and IL-6 are higher than the predetermined thresholds for CXCL10 and IL-6, respectively. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the levels of CCL2 and IL-6 are higher than the predetermined thresholds for CCL2 and IL-6, respectively. In some embodiments, a subject can be selected for treatment with a TYK2 inhibitor when the levels of CXCL10, CCL2, and IL-6 are higher than the predetermined thresholds for CXCL10, CCL2 and IL-6, respectively. TYK2 Inhibitors Drug repurposing accelerates testing therapeutic hypotheses expeditiously in clinical trials, and FDA approved drugs can serve as chemical biology probes to elucidate novel targets. By elucidating the mechanism of action of baricitinib and ruxolitinib, which target the JAK kinase family which lies downstream of TBK1 signaling, in human neural cells, TYK2 was identified in the present disclosure as a target to rescue cdsRNA-induced toxicity. Deucravacitinib reverses dsRNA-induced toxicity at lower doses than either ruxolitinib or baricitinib. Its high selectivity for TYK2 could achieve improved safety profiles relative to baricitinib and ruxolitinib. The inhibition of JAKI-3 is associated with adverse infectious, embolic, and thrombotic, neoplastic, and gastrointestinal perforation events77, which have not been reported in people with TYK2 polymorphisms78,79 or those treated with a potent TYK2 inhibitor80. Therapeutic interventions that target TYK2 to inhibit cdsRNA-induced neuroinflammation may have potential not only in AD but also in other neurodegenerative diseases with TDP-43 pathology. Methods provided herein comprise treating a subject with at least one TYK2 inhibitor after the subject has been subjected to a smell test according to the methods of administering said test disclosed herein. In some embodiments, a subject selected for treatment with a TYK2 inhibitor can be administered a therapeutically effective amount of at least one TYK2 inhibitor. In some embodiments, a subject selected for treatment with a TYK2 inhibitor can be administered a therapeutically effective amount of a selective TYK2 inhibitor. In some embodiments, a subject selected for treatment with a TYK2 inhibitor can be administered a therapeutically effective amount of baricitinib, ruxolitinib, deucravacitinib, or any combination thereof. In some embodiments, a subject selected for treatment with a TYK2 inhibitor can be administered a therapeutically effective amount of deucravacitinib. Deucravacitinib is also known as 6-(cyclopropanecarboxamido)-4-((2-methoxy-3-(l-methyl-lH-l,2,4-triazol-3-yljphenyl) amino)-N-(methyl-d3)pyridazine-3-carboxamide, having the structure of Formula (I): Other TYK2 inhibitors suitable for use herein include, for example, TYK2 inhibitors as described in WO 2012 / 000970, WO 2012 / 035039, WO 2013 / 174895, WO 2015 / 091584, WO 2015 / 032423, WO 2017 / 040757, WO 2018 / 071794, WO 2018 / 075937, WO 2019 / 023468, US 2015 / 0045349, US 2015 / 0094296, and US 2016 / 0159773, the contents of each of which are hereby incorporated by reference in their entirety herein. Methods of Treatment and Treatment Outcomes The present disclosure provides methods of treating a subject for dementia. In some embodiments, a method of treating a subject for dementia comprises assessing that the subject is at high risk for dementia as determined by the smell test according to the methods disclosed herein. In some embodiments, a method of treating a subject for dementia comprises assessing that the subject is at high risk for dementia as determined by measuring the level of least one biomarker for dementia (e.g., CXCL10, CCL2, and / or IL-6) according to the methods disclosed herein. In some embodiments, a subject that is identified as being at high risk for dementia according to the methods disclosed herein is administered at least one pharmacologic therapy for treating dementia / dementia associated with Alzheimer’s disease. In some embodiments, the method of treatment comprises administering to the subject at least one treatment for Alzheimer’s disease. In some embodiments, the method of treatment comprises administering to the subject benzgalantamine, donepezil, galantamine, rivastigmine, memantine, lecanemab, donanemab, suvorexant, brexpiprazole, or any combination thereof. In some embodiments, a subject that is identified as being at high risk for dementia according to the methods disclosed herein is administered at least one non-pharmacologic therapy for treating dementia / dementia associated with Alzheimer’s disease. In some embodiments, the method of treatment comprises administering to the subject a medical device. Non-limiting examples of medical devices suitable for use herein include deep brain stimulation (DBS), transcranial stimulation (tCS), continuous positive airway Pressure (CPAP), electroconvulsive therapy (ECT), low-energy infrared / laser LED light (IRL), photobiomodulation (PBM), transcutaneous vagal nerve stimluation (TVNS), hyperbaric oxygen chamber, low intensity pulsed ultrasound (LIPU), hearing aid placement, or any combination thereof. In some embodiments, the method of treatment comprises administering to the subject cognitive retraining. In some embodiments, the method of treatment comprises modification of a subject’s overall diet, administering specific foods, administering at least one vitamin, administering at least one vitamin minerals, or any combination thereof. In some embodiments, the method of treatment comprises administering to the subject an exercise regimen. Non-pharmacologic therapies for treating dementia / dementia associated with Alzheimer’s disease are known in the art (e.g., Olazaran et al., Dementia and Geriatric Cognitive Disorders 30.2 (2010): 161-178; and Xiao et al., Sci Rep. 2024 Apr 15; 14(1):8693) and are suitable for use herein. Methods of treating a subject for dementia as disclosed herein can result in an improvement in at least one symptom of the target disease (e.g., dementia / dementia associated with Alzheimer’s disease) following administration. Non-limiting symptoms of dementia / dementia associated with Alzheimer’s disease can include: memory loss that disrupts daily life; challenges in planning or solving problems; difficulty completing familiar tasks at home, at work, or at leisure; confusion with time or place; trouble understanding visual images and spatial relations; problems with words in speaking or writing; misplacing things and not being able to retrace steps; decreased or poor judgment; and withdrawal from work or social activities. Methods of treating a subject for dementia as disclosed herein can prevent progression of the target disease (e.g., dementia / dementia associated with Alzheimer’s disease) following administration. In some embodiments, a method of treating a subject for dementia as disclosed herein results in a change in the mini-mental state examination (MMSE) score compared to baseline. In the context of an assessment of the effect of a method of treatment, the term “baseline” refers to a measurement pretreatment. The MMSE score measures overall Alzheimer’s disease symptoms (see, e.g., Podhoma et al., Alzheimers Res Ther (2016) Feb 12;8:8). In some embodiments, the MMSE score increases in a subject treated in accordance with a method provided herein compared to baseline, indicating an improvement of symptoms. In other embodiments, the MMSE score remains unchanged in a subject treated in accordance with a method provided herein compared to baseline. In some embodiments, a method of treating a subject for dementia as disclosed herein results in a change in the Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) score (e.g., ADAS-Cog 3, ADAS-Cog 5, the original ADAS-Cog 11, and / or ADAS-Cog 13 scores) compared to baseline. ADAS-cog tests cognitive performance and has an upper limit is 85 (poor performance) and lower limit is zero (best performance) (see, e.g., Podhoma et al., Alzheimers Res Ther (2016) Feb 12;8:8; Cohen et al., JPrev Alzheimers Dis. 2022;9(3):507-522. In some embodiments, the ADAS-cog score decreases in a subject treated in accordance with a method provided herein compared to baseline, indicating an improvement of symptoms. In other embodiments, the ADAS-cog score remains unchanged in a subject treated in accordance with a method provided herein compared to baseline. Methods of treating a subject for dementia as disclosed herein can of prevent progression of a pathology associated with a neurodegenerative disease (e.g., dementia / dementia associated with Alzheimer’s disease) following administration. In some embodiments, a method of treating a subject for dementia as disclosed herein can prevent further accumulation of amyloid-beta (PA), tau proteins, intraneuronal neurofibrillary tangles (NFTs), or a combination thereof (see, e.g., Mehta et al., Clin Geriatr Med. 2023 Feb;39(l):91 -107). In some embodiments, a method of treating a subject for dementia as disclosed herein can prevent progression of a TDP-43 pathology. In some embodiments, a method of treating a subject for dementia as disclosed herein can prevent progression of cerebral atrophy. In some embodiments, a method of treating a subject for dementia as disclosed herein can prevent progression of atrophy of the hippocampus. In some embodiments, a method of treating a subject for dementia as disclosed herein can prevent progression of atrophy of the amygdala. EXAMPLES The invention is further described in the following examples, which do not limit the scope of the invention described in the claims. Materials and Methods The following materials and methods were used in the Examples below. Development and quality control of the AROMHA brain health test The ABHT was updated for remote at-home self-administration of previously developed Odor Percept Identification (OPID), Percepts of Odor Episodic Memory (POEM), and Odor Discrimination (OD) subtests

[43] , All pre-screening, informed consent, and test administration occurred online through a web-based interface (testyourbrainhealth.com). This updated version of the test consisted of five different 8.5 inch x 11 inch single-use cards packaged in one envelope and mailed to the participant’s home. We expanded the manufacturing of odor labels from the three odors utilized in the COVID smell test

[62] to include additional 15 odors. Odor labels were manufactured by MFR Samplings using Living Library™ odors purchased from International Flavors and Fragrances (IFF)

[62] , Odors were presented to participants in a peel-and-sniff manner and contained proprietary naturalistic odors from the Living Library developed by IFF (iff.com). Gas chromatography / mass spectrometry (GC / M) were conducted at the Mass Spectrometry core at the Bauer Laboratory in the Harvard Chemistry Department. Briefly, each odor label was completely opened in a stoppered 15 milliliter (ml) conical tube and allowed to reach equilibrium for 1 minute (min) at room temperature. Then a Hamilton Syringe was used to inject a representative sample of the headspace into the GC / MS instrument. The peaks were normalized to 2-methyl-3-heptanone equivalents, and analyzed for common set of peaks that might represent a common contaminant from the adhesive. All samples were run the same day to eliminate batch effects (FIGS. 3A-3C). Participant responses to all components of the ABHT were collected on a webbased application at testyourbrainhealth.com designed for independent selfadministration of the survey questions and the olfactory battery (FIGS. 4A-4D). The data was stored on a HIPPA-compliant AWS server. The prescreening module and the informed consent module were developed on a RedCAP platform at the Massachusetts General Hospital. All protected health information was kept on RedCAP platform. Participants had the option to call a research assistant for live help in English or Spanish at any time during remote testing. The web-based AHBT application directed participants to a RedCap secure e-consent project to collect identifiers. Once consented, participants were sent back to the AROMHA, Inc. webbased application to walk through all three parts of the battery and collect olfactory information associated with their card ID. The web-based application was designed to lead participants through every stage of testing, including directions on how to peel odor labels and sample odors as well as how to respond to questions regarding odor intensity, odor identification & naming confidence, odor memory, and odor discrimination. The web-based application had the ability to run in an English or Spanish language mode, based on participant preference. During testing, the webbased application collected participant responses for all aspects of the olfactory battery in addition to the timing of those inputs. The web-based application generated summary and item-specific data on these metrics that was downloaded by researchers for analysis and joined offline to demographic information collected in RedCap following the e-consent process. These results were not shared with participants. Part 1: Odor Percept Identification Test (OPID9). Participants first completed the OPID9, which involved identifying nine distinct odors: menthol, clove, leather, strawberry, lilac, pineapple, smoke, soap, and grape. These odors were selected for their predictive value in identifying the conversion to Alzheimer’s disease (AD) in patients with mild cognitive impairment (MCI)

[39] , After each odor presentation, participants rated the intensity on a Likert scale from 0 to 10. Subsequently, participants were presented with four odor names, asked to choose the label that best represented the odor they sampled, and asked to rate their confidence in their identification choice using the following scale: “I Guessed,” “I Narrowed Down to Three,” “I Narrowed Down to Two,” or “I Am Certain.” (FIG. 4B). This process was repeated for all nine odors. Following the completion of the OPID9 test, there was a 10-minute delay during which participants answered demographic, medical, nasal, and memory-related questions. Part 2: Percepts of Odor Episodic Memory (POEM) / OPID18. After the 10-minute break, participants completed the POEM / OPID18 tests. These tests included the nine odors from Part 1 and nine additional odors: coffee, peach, chocolate, orange, dirt, banana, lemon, bubble gum, and rose. The odors were presented in a stereotyped random order that was held consistent across all participants. For each odor, participants first indicated whether the odor sampled was presented in Part 1 (yes / no), OPID9. As in the earlier odor identification test, participants then selected the odor name most representative of the odor from four choices and rated their confidence in their selection. Part 3: Odor Discrimination (OD10). In Part 3, participants were presented with 10 pairs of odors, all of which were previously presented in Parts 1 and 2. Participants were asked to determine whether the paired odors were the same or different (yes / no). The POEM index was calculated as the difference between the proportion of correct and incorrect recognitions, with scores ranging from -1 to 1. OPID9 and OPID18 scores were calculated as the total number of correctly identified odors, with maximum scores of 9 and 18, respectively. The OD10 score was the total number of correctly discriminated odor pairs, with a maximum score of 10. The average intensity score was derived from the mean intensity ratings of the nine odors from Part 1 on the Likert scale. OPID9noguess and OPID18noguess scores were calculated as the total number of odors identified correctly where the participant did not select “I Guessed” for the confidence question immediately following identification. After verifying participants’ comfort with using the web-based application in conjunction with the AROMHA Brain Health Test’s smell cards through the use of research assistant observation, cognitively healthy participants were given the option to complete the test with or without the observation of a research assistant. Verified cognitively impaired participants were not given this option and were only able to self-administer the smell test remotely or in person under the observation of a research assistant. Participants In the examples, 127 age-matched control subjects (CN), 34 subjects with Subjective Cognitive Concerns (SCC), and 19 subjects with Mild Cognitive Impairment (MCI) participated in the study. The cognitive status classification was verified for subjects recruited (n=59) using the National Alzheimer’s Coordinating Center (NACC) uniform dataset procedure

[99] , Cognitive status classifications (CN, SCC, MCI) were made clinically, either by a consensus committee of expert clinicians (e.g., neurologists, neuropsychologists, geriatricians) or by one or more clinicians. The decision was based on (1) the presence of concern about cognitive change raised by the subject, a co-participant, or a clinician, (2) the presence / absence of impairment in one or more cognitive domains, and (3) the independence in functional abilities. Demented participants were excluded from the study. The severity of cognitive and functional impairment was measured with the Clinical Dementia Rating (CDR) scale; cognition was assessed using the neuropsychological battery from the National Alzheimer’s Coordinating Center (NACC) uniformed data set (UDS). This battery included a cognitive screening (Montreal Cognitive Assessment - MoCA), and assessments of memory (Immediate and Delayed Recall from Logical Memory or Craft Story), attention / working memory (Forward and Backward Digit Span or Number Span), processing speed / executive functioning (Trail Making Test, Parts A and B), language (Category Fluency, Boston Naming Test or Multilingual Naming Test), and visuospatial skills (Benson Complex Figure Copy). Unverified participants recruited were classified cognitively based on the presence or absence of self-reported cognitive complaints (SCC) or medical diagnoses affecting cognitive function (MCI) (n=l 18). All participants underwent informed consent before participation. The research protocol was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Partners Health. Unverified MCI participants (n=4) were aged 55+ and reported a clinical diagnosis of MCI by a certified physician. Verified MCI participants (n=15) were aged 55+, met the National Institute on Aging-Alzheimer’s Association (NIA-AA) diagnostic criteria for MCI

[100] , with a performance below the demographic / education-adjusted norm on neuropsychological assessment indicating MCI, had a CDR score from 0.5 to 1 and had preserved IADL (determined by a clinician), and were not demented. In total, 19 participants with MCI were recruited. Unverified Subjective Cognitive Concern (SCC) participants (n=13) were aged 55+ and reported SCC based on the following 3 questions: (1) Have you experienced a change in your memory in the last 1-3 years?; (2) Has this been a persistent change over the last 6 months?; and (3) Are you concerned about this change? Responses were on a Likert scale: “Not at all,” “Slightly,” “Moderately,” “Considerably,” and “Extremely.” If participants endorsed “Slightly” or worse to all 3 questions, they were categorized as SCC. Verified SCC participants (n=18) reported significant SCC based on the above three questions while not meeting MCI criteria. Three participants under 55 years old reported SCC and were removed from the analyses. In total, 31 participants with SCC were included. Unverified CN (n=99) were recruited from the internet and were aged 18 years old and over and did not report any cognitive complaints or medical diagnoses affecting cognitive function. From the unverified CN participants, subgroup of unverified CN aged 55+ (n=44) was created. Verified CN aged 55+ recruited (n=29), had a normal cognition (CDR=0 and / or neuropsychological testing within normal range), did not report subjective memory concerns based on the three questions presented above, and did not meet MCI criteria. It was ensured that demographic variables were similar across verified and unverified CN older adults (control group). T-test and chi-square analysis revealed no statistical difference across modalities for age (p=0.47), sex (p=0.99), and education (p=0.27). Wilcoxon rank sum / Mann Whitney tests were performed to compare the median performance on each olfactory measure between verified and unverified participants in each subgroup (CN, SCC, and MCI). Since, after adjustment for multiple comparisons, no difference was significant at a threshold of 0.05, the verified and unverified participants were combined within a subgroup for the purposes of these analyses. Clinical data comparing the verified participants in each group are 5 presented and compared in Table 1. Table 1: Clinical values and olfactory function across cognitive status in verified older adult participants. p values Verified CN older adults (n=28) Verified SCC (n=18) Verified MCI (n=13) CN vs. SCC CN vs. MCI SCC vs. MCI Age (years) 72.75 (10.96) 78.17(9.31) 77.69 (5.31) .08 .06 .86 Sex (female %) 61% 56% 54% .97 .94 .99 Education (years) 15.71 (4.57) 16.11 (2.56) 16.08 (3.04) .71 .77 .97 Clinical Dementia Rating (CDR) 0.00 (0.00) 0.42 (0.19) 0.54 (0.14) <.001 <.001 .049 MoCA (z-score adjusted for age, sex, and education) 0.29 (0.84) -0.24(1.17) -0.84 (1.06) .11 .003 .15 OPID9 ( / 9) 6.07(1.21) 5.11 (1.64) 3.00 (1.78) .04 <.001 .049 OPID9noguess ( / 9) 5.29 (3.67) 4.29(1.79) 3.16(1.77) .008 .001 .32 OPID18 ( / 18) 11.25 (2.74) 10.00 (3.09) 7.31 (2.95) .08 <.001 .04 OPID18noguess ( / 18) 9.46 (3.82) 7.00 (3.79) 4.77 (3.32) .04 <.001 .09 OD10 ( / 10) 8.29 (1.49) 7.39(1.69) 7.15 (2.12) .07 .10 .74 POEM (-1 to +1) 0.24 (0.27) 0.23 (0.25) 0.25 (0.2) .96 .83 .80 Average intensity ( / 10) 5.29 (2.11) 4.71 (1.40) 4.78 (2.30) .27 .50 .93 Test Duration (minutes) 41.99 (20.88) 42.26 (15.28) 39.91 (8.64) .96 .65 .59 Note: Values are means (SD). T-tests were performed across groups for age, education, clinical variables, olfactory scores, and test duration. Chi-square test was performed on sex proportion across groups. Bold indicates a statistically significant difference that remained significant after Bonferroni correction for 21 comparisons (p < .002). One participant from the Verified CN older adults ’ group was excluded due to missing data. Statistical analysis (AROMHA brain health test) 10            All analyses were conducted in R using Base R, effectsize, emmeans, and rstatix packages. Student T-tests were used to compare demographics and olfactory variables between observed and unobserved conditions and between English and Spanish-speaking groups, while the Wilcoxon signed-rank test was used to compare anosmic patients with CN participants. To assess the effect of age on olfaction, linear regression models were conducted with age and olfactory scores. This analysis was performed for each olfactory subtests (OPID9, OPID9noguess, OPID18, OPID18noguess, OD10, POEM, and average intensity). To assess the effect of cognitive status on olfaction, ANCOVAs were performed to compare olfactory functioning (OPID9, OPID9noguess, OPID18, OPID18noguess, OD10, POEM, and average intensity) among older adults without cognitive impairment, participants with SCC, and with MCI, including age, sex, and education as covariates. For all the analyses, the alpha value was set at 0.05 and Bonferroni correction was used for multiple comparisons. Histological Staining and Imaging All human tissues were acquired from the MGH MADRC brain bank in accordance with protocols approved by the MGB Institutional Review Board. Similar to what was previously reported4, ~5 pm thick formalin-fixed paraffin-embedded (FFPE) sections of the amygdala were dewaxed followed by antigen retrieval in a Leica Bond Fully Automated Slide Stainer. After deparaffination and rehydration by sequential incubation in Leica dewing solution (3x5 min), 100% ethanol (2x5 min), 90% ethanol (2x5 min), 70% ethanol (2x5 min), and distilled water (3x5 min), antigen retrieval was achieved with citrate-based antigen retrieval solution (ER2) and steam cooking for 1 hour. Afterwards, sections were cooled to room temperature and rinsed in water and lx PBS. For permeabilization, 0.03% hydrogen peroxide solution was applied for 8 min. After blocking in tris-buffered saline (TBS)-based Odyssey blocking buffer (cat. # 927-60001, LLCOR, NE, USA) for 1 hour, for immunohistochemistry, primary antibodies (J2 for dsRNA (cat. # 76651, Cell Signaling Technology, MA, USA); pTDP-43 S409 / 410 (cat. # CAC-TIP-PTD-P02, Cosmo Bio, CA, USA); tau (cat. # NBP2-25162, Novus Biologicals, CO, USA); AB (cat. # CST-D54D2, Cell Signaling Technology)) diluted in blocking buffer were applied overnight at 4°C in a humidified chamber. Following a wash with PBS (3 x 5 min), secondary antibodies were applied using the VECTASTAIN ABC Staining Kit (cat. # PK-6200, Vector Laboratories, CA, USA) and DAB Staining Kit (cat. # SK-4100, Vector Laboratories). For luxol fast blue-hematoxylin and eosin (LHE) staining, the protocol of the Luxol Fast Blue Stain Kit (cat. # ab 150675, abeam, MA, USA) and the Sakura Prisma H&E Stain Kit (cat. # 76318-76, VWR, PA, USA) were followed. For tissue-based cyclic immunofluorescence (CyCIF)34, unspecific binding sites were blocked, slides were incubated in 3% H2O2 and 20 mM HC1 in PBS at room temperature with light-emitting diode illumination for 1 hour to reduce autofluorescence, and the first primary antibody set (J2 and pTDP-43 S409 / 410) was applied. After washing, the secondary antibodies (anti-rabbit Alexa Flour 555 (cat. # A-31572, Thermo Fisher, MA, USA); anti-mouse Alexa Flour 647 (cat. # A-21237, Thermo Fisher)) 1:1000 diluted in blocking buffer were applied for 1 hour in a humidified chamber at room temperature. After a wash, Sytox Blue Nucleic Acid Stain (cat. # SI 1348, Thermo Fisher) was applied 1:15000 in blocking buffer for 15 min at room temperature. Following a final washing step, slides were imaged. After this first cycle, the fluorophores were inactivated by 4.5% H2O2 and 24 mM NaOH in PBS at room temperature under light exposure for 1 hour. After rinsing, the second primary antibody set (tau and AB) was applied and staining was continued in the same manner as described above. Differential Gene Expression Analysis of Different Brain Regions Differential Gene Expression was conducted based on a previously established protocol91. In brief, clinical diagnosis was used to define AD versus control conditions within the bulk RNA-sequencing data taken from the ROSMAP37 and MSBB38 databases (data obtained from the AMP-AD Knowledge Portal (doi:10.7303 / syn2580853)). The analysis focused on a list of 354 pre-selected interferon-related genes. Of these genes, 309 could be found in the analyzed brain regions (from ROSMAP: dorsolateral prefrontal cortex (DFPC), from MSBB: frontal pole (FP), inferior frontal gyrus (IFG), parahippocampal gyrus (PHG), superior temporal gyrus (STG)). 139 genes were significantly upregulated (adjusted P value < 0.05) while 51 were downregulated. Induced Pluripotent Stem Cell (iPSC) differentiation and treatment An isogenic pair of a recently established iPSC line that harbors a pathogenic TDP-43 mutation (TDP-43+ / G298S) in the FA0000011 genetic background35,92 was differentiated into NGN2 cortical-like neurons using a recently published protocol36,93 in a 384-well plate format (10,000 cells per well). Where indicated, wildtype FA0000011 cells were transfected with poly(I:C) HMW (cat. # tlrl-pic, Invitrogen, CA, USA) by lipofection with lipofectamine 2000 (cat. # 11668027, Thermo Fisher) (final dsRNA concentration: 4-8 pg / ml, depending on poly(I:C) batch effects). ReNcell VM and CX cell culturing and differentiation ReN VM (cat. # SCC008, Millipore, MA, USA) and CX cells (cat. # SCC007, Millipore) were cultured in ReNcell Maintenance media (cat. # SCM005, Millipore) containing 1:100 of Penicillin-Streptomycin (cat. #30-002-CI, Corning, NY, USA), 20 ng / ml of epidermal growth factor (cat. # GF001, Millipore) and 20 ng / ml of basic fibroblast growth factor (cat. # 03-002, Stemgent, MD, USA) and differentiated into neural cells for one week by removing the growth factors as previously described4,94, with the difference that the differentiation took place in a separate dish (FIGS. 20A-20B) to minimize variability downstream. SH-SY5Y culturing SH-SY5Y Neuroblastoma cells (cat. # CRL-2266, ATCC, VA, USA) were cultured in 1:1 Eagle’s Minimum Essential Medium (EMEM) (cat. # 10-009-CV, Corning) andF12 medium (cat. # 11765-047, Thermo Fisher) containing 1:100 of Penicillin-Streptomycin (cat. #30-002-CI, Coming) and 10% fetal bovine serum (FBS; cat. # 10438-026, Thermo Fisher). Prior to any treatments, cell media was changed to media containing 1% FBS instead of 10% to slow down cell division. ReN VM, CX and SH-SY5Y cell treatments SH-SY5Y cells as well as differentiated ReN VM and CX cells were seeded into a 96-well plates (VM / CX: 20,000 cells per well with Matrigel-coating (cat. # 354320, Corning), SH-SY5Y cells: 10,000 cells per well without coating) and where applicable, respective drugs dissolved in dimethyl sulfoxide (DMSO) (baricitinib: HMS LINCS ID # 10354, ruxolitinib: HMS LINCS ID # 10138, deucravacitinib: cat. # HY-117287, MedChem Express, NJ, USA, MG-132: cat. # HY-13259, MedChem Express, etoposide: cat. # HY-13629, MedChem Express, menadione: cat. # M9429-25G, Sigma-Aldrich, MO, USA) were added using a D300e Digital Dispenser (Hewlett-Packard, CA, USA). After 1 hour of incubation at 37 °C, cells were transfected with poly(I:C) HMW where applicable (cat. # tlrl-pic, Invitrogen) by lipofection with lipofectamine 2000 (cat. # 11668027, Thermo Fisher) (final dsRNA concentration: 4-8 pg / ml, depending on poly(I:C) batch effects). To test the relevance of IFNAR2, a neutralizing anti-IFNAR2 antibody (cat. # 21385-1, PBL Assay Science, NJ, USA) was added in the respective amount to each well after lipofection and without prior drug treatment. To test whether interferon a and poly(I:C) have comparable effects on neurons, interferon a (cat. # 407294-5MU, Millipore) was diluted in media and added to the cells without prior drug treatment. Western blot samples and cell media for biomarker testing were collected 24 h (48 h for MG-132 treatment) post-treatment. Cell viability was assessed 72 hours (VM, SH-SY5Y) and 7 days (CX) post-treatment using the CellTiter-Glo Assay (cat. # G7572, Promega, WI, USA) following the manufacturer’s instructions. All conditions were normalized to 0.1% DMSO content. Where applicable, lipofectamine 2000 (VM, CX) or lipofectamine 2000 + 10 pM drug (SH-SY5Y) was used as the respective control. For mass spectrometry, two million ReN VM cells were expended in a 15-cm-dish (2 dishes per replicate) for 4 days and subsequently differentiated for 1 week, then treated with lOpM of drug, transfected with poly(I:C) after 1 hour of incubation at 37 °C, and 24 hours later, washed with ice-cold lx PBS and collected by scraping followed by centrifugation at 500 g for 5 min. For the testing of common stressors, time-lapse images were captured every 3 hours for 33 hours using GE In-Cell 6000 Analyzer (GE Healthcare, IL, USA) under the confocal mode. Each image's total cell surface area and average fluorescence intensity were quantified using Fiji95 and plotted using GraphPad Prism. Four biological repeats were taken for each group. Cell Staining and Imaging Differentiated ReN VM cells treated with 1 pM MG-132 were fixed 48 h posttreatment and iPSC-derived MGM2 cortical-like neurons were fixed after ten days of differentiation using 4% paraformaldehyde (cat. # 28908, Thermo Fisher) at room temperature and protected from light. Additionally, after three washes with 1 x PBS, ReN VM cells were permeabilized using ice-cold methanol for 10 min. Cells were washed gently with 1 x PBS three times and then incubated with blocking solution (cat. # 927-60001, LLCOR) for 1 hour at room temperature. All blocking solution was removed, and cells were incubated with fresh blocking solution and primary antibodies overnight on a shaker at 4 °C. Primary antibodies used were anti-TUJl (1:250, cat. # TUJ, Aves Lab, CA, USA), KI anti-dsRNA (1:250, cat. # 10020200, Scicons, Susteren, The Netherlands), and anti-TDP-43 (1:100, cat. # 3448, Cell Signaling Technology). For poly(I:C) antibody incubation controls, poly(I:C) was incubated with the dsRNA primary antibodies at a ratio of 5x poly(I:C) : lx dsRNA primary antibody for 1 hour before being added to cells. Three PBS washes were done, and cells were incubated with secondary antibodies based on species at a 1:500 (iPSCs) and 1:1000 (ReN VM) dilution in the blocking solution for 1 hour at room temperature on a shaker. Hoechst 33342 (cat. # H3570, Thermo Fisher) was added at a 1:1000 dilution to the well with iPSCs for the final ten minutes. For ReN VM cells, cells were first washed three times with 1 x PBS and then Hoechst was added to the wells in a 1:5000 dilution in PBS for 30 min at room temperature. Secondaries used were donkey anti-chicken Alexa Fluor 488 (cat. # 703-545-155, Jackson ImmunoResearch, PA, USA), donkey anti-rabbit Alexa Fluor 568 (cat. # Al 0042, Invitrogen), donkey anti-mouse Alexa Fluor 647 (cat. # A31571, Invitrogen), and donkey anti-rabbit Alexa Fluor 488 (cat. # A21206, Invitrogen). After three final washes with 1 x PBS, Confocal images were acquired with an Image X-Press Micro Confocal (Molecular Devices, CA, USA). Images were taken with a 20X (ReN VM) or 40X (iPSC) objective with four z-stacks (1.5 pM step size) per field for between 25 and 55 fields per well. All groups were imaged with the same acquisition settings within each experiment. Automatic quantifications were performed using a custom Fiji / ImageJ-based plugin (National Institutes of Health, version 1.53c)95 that measures integrated intensity of KI dsRNA antibody within TUJ1-positive cells. RNA-sequencing data processing and prediction of TDP-43 status RNA-sequencing reads of brain specimens from Alzheimer’s disease patients in the ROSMAP study, along with their corresponding clinical annotations, were downloaded from the AMP-AD Synapse portal at adknowledgeportal.synapse.org. RNA-sequencing reads from ALS patient neurons that were sorted for presence or absence of TDP-43 were downloaded from Sequence Read Archive (SRA) accession GSE12654243. Transcripts were quantified using Salmon vl.9.096 against release 107 of the hg38 human transcriptome from Ensembl. The transcriptome was amended with three transcripts that have not yet been annotated in Ensembl, but that are known to be associated with loss of nuclear TDP-4318. Specifically, we included two splice variants of UNCI 3 A, each including an additional CE between the canonical exons 20 and 21 (hg38; chrl9: 17,642,414-17,642,541 (CE1); chrl9: 17,642,41417,642,591 (CE2)), and one splice variant of STMN2, which included an alternative exon 2 termed exon 2a (hg38; chr8:79,616,822-79,617,048). Given our quantification of TDP-43 pathology associated transcripts, we classified patient samples as TDP-43-positive or -negative based on the number of expressed CEs. We dichotomized transcript abundance using thresholds set above the lowest non-zero peak in the abundance histograms (TPM>1.3 STMN2short; TPM>0.09 for UNC13A-CE1; TPM>0.08 UNC13-CE2). This threshold determined whether each individual CE was present or absent. Next, patient samples expressing one or none of the three CEs (STMN short, UNC13A-CE1 or UNC13A-CE2) were considered TDP-43-negative and patient samples expressing two or all three of the CEs were considered TDP-43-positive. Using this classification, the proportion of TDP-43 pathology positive samples was approximately 37%. Our classification was a conservative prediction, given that the clinical incidence of TDP-43 inclusions in AD patients was up to 57%14. Prediction of disease stage and assessing gene set significance For assessing how well any given gene set predicts Alzheimer’s disease progression we used an updated variant of the published DRIAD-SP methodology26. For predicting disease stage, we collapsed transcript abundances to gene-level abundances using tximport97. The entire transcriptional feature space was filtered down to the about ~20 k protein-coding genes in the human genome to ensure that only genes with gene products that can conceivably be targeted using conventional small molecule inhibitors are included. Every AMP-AD specimen was assigned a label of disease severity based on the following mapping to the Braak annotations: A—early (Braak 1-2), B—intermediate (Braak 3-4), and C—late (Braak 5-6). We set up an ordinal regression task classifying patients according to their disease stage using the ordinalRidge R package (github.com / labsyspharm / ordinalRidge). We used the gene expression measurements from AMP-AD as predictors, filtered to only contain genes in the current gene set under investigation. This way, the performance (AUC) of a model utilizing a given gene set could be viewed as a measure of how strongly the given genes are predictive of AD progression. To address overfitting, ordinalRidge utilized a ridge regularization term that penalized the L2-norm of feature weights. No LASSO regularization was used, as it induced sparsity and excluded features that were specifically preselected to be included in the model. Model performance was evaluated through leave-pair-out cross-validation. For a given regression task, each example in the dataset was associated with an example from the other classes that was the closest match in age. If there were multiple candidates for the age match, the pairing was selected uniformly at random. The resulting set of age-matched pairs was evaluated in a standard cross validation setting by asking whether the later-stage example in each withheld pair was correctly assigned a higher score by the corresponding predictor. The fraction of correctly ranked pairs constituted an estimate of the area under the ROC curve98. Comparing model performances across gene sets directly was not possible, as their set sizes can differ. Therefore, we normalized AUC values by relating them to performances observed on size-matched random gene sets. For each gene set of interest, 1000 random gene sets of matching sizes were sampled from a uniform distribution over the protein-coding space. After evaluating all lists through crossvalidation, an empirical P value was computed as the fraction of background sets that yield higher predictor performance than the gene list of interest. CRISPR screen Low passage (P2) ReNcell VM cells stably expressing Cas9 were expanded and transduced with lentiviral particles containing the Brunello library50. Lentiviral transduction efficiency was tested by plating cells into two 6-well plates and varying volumes of stock preparation of Brunello library lentiviral vectors with 8 pg / ml of polybrene. One plate was treated with puromycin, the other plate remained untreated, and the proportions of cells measured for each condition using CellTiter-Glo were used to calculate the efficiency which was found to be at 84% averaged from triplicates. Transduced cells were expanded in the presence of puromycin until the required number of cells was achieved to obtain triplicates of 1000 cells per sgRNA. Then, cells were differentiated as described above. After seven days, 3 pg / ml dsRNA in complex with lipofectamine (concentration according to manufacturer’s protocol) or lipofectamine alone as a control were added to each flask. After 3 days of treatment, genomic DNA was isolated from each group of cells. PCR and sequencing were performed as previously described99,100. For analysis, the read counts were normalized to reads per million and then log2 transformed and the LFC of treatment and control were compared to the lipofectamine controls. Mass spectrometry First, cells were lysed in lysis buffer (2% SDS, 150 mM NaCl, 50 mM Tris pH 8.5) supplemented with protease and phosphatase inhibitor (cat. # 11873580001 and 4906845001, Millipore) and homogenized with a QIAshredder column (cat. # 79656, Qiagen, Hilden, Germany). Lysates were reduced with freshly prepared dithiothreitol (DTT) (final concentration: 5 mM) and heated at 37 °C for 1 hour. Samples were then alkylated with iodoacetamide (final concentration^© mM) for 25 min with 50 mM ammonium bicarbonate in the dark whereupon the reaction was stopped by adding DTT (final concentration: 50 mM). After methanol / chloroform precipitation, the purified protein was solubilized in freshly prepared 8 M urea in 200 mM EPPS, at pH 8.5 for 30 mins at 37 °C. For digestion, the urea concentration was diluted with 200 mM EPPS to 4 M final concentration and 2% acetonitrile (v / v) was added. Digestion was performed using lysyl endopeptidase® Lys-C (2 mg / ml; cat. # 121-05063; Wako, Osaka, Japan), (enzyme-to-substrate ratio: 1:50) for 3 hours at 37 °C. Following dilution of the sample with 200 mM EPPS to a final urea concentration of 1.6 M, further digestion was performed by the addition of trypsin (cat. # V5111, Promega), (enzyme-to-substrate ratio: 1:100) for 8 hours at 37 °C. A small aliquot was analyzed for its missed cleavage rate by mass spectrometry. Equal amounts of digested protein (60 pg) were removed from each sample and labeled using TMTpro 16plex Mass Tag Labelling Kit (TMT; cat. # A44520, Thermo Fisher) following the manufacturer’s instructions. Labeling efficiency of > 95% was determined along with ratio checks by mass spectrometry MS3, while the labeling reactions were stored at -80 °C. To quench the reaction, hydroxylamine was added to a final concentration of 0.5% (v / v) for 10 mins. After acidification using formic acid, equal amounts of labelled peptide were pooled from each sample (as judged from ratio check data). Then, solvent was evaporated. The labeled peptide multiplex was desalted by solid-phase extraction (SPE; SepPak tC18 Vac RC Cartridge). The peptides were fractionated by HPLC alkaline reverse phase chromatography (Agilent 1200 Series) into 96 fractions and combined into 24 samples. Lastly, peptides were desalted over Stage Tips101 prior to MS analysis. A MultiNotch SPS-MS3 TMT method102 was used on an Orbitrap Lumos mass spectrometer (Thermo Fisher) coupled to a Proxeon EASY-nLC 1200 liquid chromatography (LC) system (Thermo Fisher). Samples were injected onto a 40 cm, 100 pm (internal diameter) column packed with 2.6 pm Accucore C18 resin (flow rate of 450 nl / min). Over the course of 4 hours, the peptide fractions were separated through acidic acetonitrile gradients by the LC before being injected into the mass spectrometer. First, an MS1 spectrum (Orbitrap analysis; resolution 120,00; mass range 400-1400 Th) was taken. Then, an MS2 spectrum was collected after collision-induced dissociation (CID, CE = 35) with a maximum ion injection time of 150 ms and an isolation window of 0.7 Da. For tandem mass tag (TMT) quantification of peptides, MS3 precursors were fragmented by high-energy collision-induced dissociation (HCD, CE = 40%) and analyzed in the Orbitrap at a resolution of 50,000 at 200 Th. See, e.g., Paulo JA et al., J Proteomics 148, 85-93 (2016). A Sequest-based in-house software was used to search peptides against a human database with a target decoy database strategy and a false discovery rate of 1%. Oxidized methionine residues (+15.9949 Da) were dynamically searched, along with static modifications for alkylated cysteines (+57.0215 Da) and the TMTpro reagents (+304.207145 Da) on lysines and the N-termini of peptides. Relative protein quantification required a summed MS3 TMT signal / noise > 200 over all TMT channels per peptide and an isolation specificity > 70% for any given peptide. Quant tables were generated and exported to Excel for further procession. More details on the TMT intensity quantification and certain parameters, see, e.g., Paulo JA et al., J Am SocMass Spectrom 27, 1620-1625 (2016). Western blotting First, 15 pl of x Laemmli Sample Buffer (cat. # 1610737; Bio-Rad Laboratories, CA, USA) with 1:20 2-mercaptoethanol was added to each well after aspiration of the media. Subsequently, the lysates were boiled at 100 °C for 5 min. Then, the samples were loaded onto a precast gel (cat. # 4569036; Bio-Rad Laboratories, CA, USA) which was run in lx running buffer diluted from a lOx Tris / Glycine / SDS stock (cat. # 1610772; Bio-Rad Laboratories) at 160 V for 40 min. The transfer was performed using a PVDF membrane (cat. # IPFL00010; Millipore) and a lx transfer buffer diluted from a lOx stock (cat. # PI35040; Thermo Fisher, MA, USA) mixed with 20% methanol. For 90 min, the gel was blotted at 4 °C at 90 V. Finally, unspecific binding sites were blocked by phosphate-buffered saline (PBS)-based Odyssey® blocking buffer (cat. # 927-40150; LI-COR) for 1 hour at room temperature. Then, primary antibody (pSTATl Y701 (cat. # 9167S; Cell Signaling Technology); beta actin (ACTB, cat. # 3700; Cell Signaling Technology) diluted 1:1000 in blocking buffer was applied overnight at 4 °C. Then, the blot was washed with tris-buffered saline (cat. # sc-262305; Santa Cruz, TX, USA) containing 0.05% Tween 20 (cat. # BP337-500; Thermo Fisher) (3 x 5min). Afterwards, secondary antibody (IRDye 800CW (cat. # 926-32211; LLCOR); IRDye 680RD (cat. # 92668070; LI-COR)) diluted 1:2500 in blocking buffer was applied for 1 hour at room temperature. After three final washes, the blot was imaged with the Odyssey® DLx Imager (LI-COR). Images were analyzed using ImageStudioLite (LI-COR). Genome-wide association tests using TYK2polymorphism (rs34536443) We used an established partial loss-of-function in TYK2 (rs34536443) to estimate the effect of pharmacological TYK2 inhibition on the circulating CXCL10 concentration55 57. Summary genetic association data was obtained from a genomewide association study (GWAS) using the Olink proteomics platform in 54,219 participants of British ancestry in the UK Biobank53. Genome-wide association tests were performed using REGENIE v.2.2.1 in a two-step procedure to account for population structure in a population of European ancestry (n = 34,557). Protein levels were inverse-rank normalized. Association models included the following covariates: age, age2, sex, age * sex, age2 * sex, batch, UK Biobank center, UK Biobank genetic array, time between blood sampling and measurement, and the first 20 genetic principal components of ancestry. Summary genetic association data on these measures were also obtained from a GWAS of these measures using the SomaScan platform in 35,559 individuals of Icelandic ancestry in DeCODE54. Rank-inverse normal transformed proteins were adjusted for age, sex, and sample age. Residuals were re-standardized using rank-inverse normal transformation and standardized values were used in genome-wide association testing using the linear mixed model implemented in BOLT-LMM105. Meso Scale Diagnostics (MSD) electrochemiluminescence (ECL) biomarker assay To measure CXCL10 (IP-10) biomarker levels in collected cell media, we used the commercially available V-PLEX assay from MSD, LLC (cat. # K151NND, MSD, Cambridge, MA, USA). We used the proposed procedures with the exception that a 1:100 dilution of cell media was required for the sample values to fall within the 0.37-500 pg / ml dynamic range of the assay. All plates were analyzed under the MESO QuickPlex SQ 120 (Model # 1300) plate reader and concentrations were quantified via the Meso Scale Discovery WorkBench. Samples with concentrations reported as “NaN” were quantified as zero, as these concentrations fell below the dynamic ranges reported above. Statistical analysis Statistical analyses were conducted on GraphPad Prism for unpaired t tests with two tails and ordinary one-way analysis of variance (ANOVA) with post hoc tests. Correction for multiple comparison was done using Dunnett’s statistical hypothesis testing. For image quantification, data were plotted and statistically analyzed on R-Studio software (Version 4.3.3). Boxplots and bars represent the mean and standard error of the mean, respectively. Differences were considered significant if P < 0.05. P value style: P > 0.05 and P = 0.05 (no asterisks), P < 0.05 (*), P < 0.01 (**), P < 0.001 (***), P < 0.0001 (****) Where necessary, an outlier test (ROUT with Q = 5% and Grubbs with a = 0.05) were carried out. No outliers were eliminated from the dataset. Example 1: Design and implementation of the AROMHA brain health test. The workflow of the self-administered ABHT included tests of odor percept identification (OPID), percepts of odor episodic memory (POEM), and odor discrimination (FIGS. 1A and IB), which paralleled previous researcher-administered tests, where odors were delivered through an olfactometer

[43] or through hand-held, repeat use devices (Whispis)

[64] , The ABHT leveraged the remote administration aspects of the CO VID Smell Test

[62] by delivering the odor stimuli using odor labels arrayed on mailable cards, by including an odor intensity measure, and by enabling self-administration by developing a web-based platform

[65] to collect responses. Additionally, the ABHT added a meta-cognition measure embedded in the odor percept identification tasks. Participants were instructed by the web-based application to sample the odor, and then choose an odor name from a forced choice list of 4 options. They were then asked to evaluate their confidence in each odor identification decision with a scale that included the following options: “I Guessed,” “I Narrowed Down to Three,” “I Narrowed Down to Two,” or “I Am Certain.”. This confidence metric was quantified for the OPID9 and OPID18 odor identification tests as the number answered correct among items paired with the “I am Certain”, “I Narrowed Down to Three”, and “I Narrowed Down to Two” responses (OPID9noguess, OPID18noguess scores). Three sets of bilingual (English / Spanish) cards, which were arrayed with odor labels, were designed to administer the OPID9 odor percept identification and odor intensity test (Part 1, Card A (FIG. IB, top panel)), the OPID 18 odor percept identification and POEM memory test (Part 2, Cards B and C (FIG. IB, middle panel)), and the 0D1O odor discrimination test (Part 3, Cards D and E (FIG. IB, bottom panel)). Numerous concentrations of each odor were packaged in different labels, and perceptions of odor intensity using a 10-point Likert scale that ranges from 0 (no odor) to 10 (strongest odor imaginable) were obtained from healthy collegeaged participants in pilot studies. The final label-embedded odorant concentrations were selected with a mean perceived intensity of 7-7.5. To ensure that the headspace for each label was not contaminated by one or more components of the adhesive to hold the labels together, gas chromatography / mass spectrometry for each odor label was performed (FIGS. 3A-3B). A common component in the headspace of all labels, which could confound olfactory performance, was not found. The foil choices were orthogonal to the target odor and evocative rather than generalized. This minimized cognitive load by offering distinct alternatives and prevented real-world contextual biases that could lead to associative errors. For example, foils describing fruit odors were not used when the target odor was a fruit. Evocative foil names like “coconut” or “fresh bread” were also incorporated. Additionally, foils with contextual associations often encountered with the target odor in real life (e.g., soap and vanilla, lavender and chamomile) were specifically avoided to further reduce potential bias. Finally, the set of odor names was expanded so that each odor name was only presented once within the OPID9 or OPID18 odor identification tests, either as a target odor or a foil (Tables 2 and 3). When the 9 odors, presented in the first OPID9 odor identification and intensity test, were presented again (after a 10-minute break) in the OPID18 odor identification and POEM memory test, new sets of foils were presented with the correct name for each target odor to vary the identification experience and reduce learning carryover from OPID9 (Tables 2 and 3). Table 2: Odors and four choices presented to participants for the OPID9 test. Trial Odor Name 1 Odor Name 2 Odor Name 3 Odor Name 4 Practice trial cherry (cereza) glue (pegamento) grass (cesped) tomato sauce (salsa de tomate) 1 lime (lima) soy sauce (salsa de soya) coconut (coco) menthol (mentol) 2 barbecue (parrilla) clove (clavo) popcorn (palomitas de maiz) olive (aceituna) 3 rosemary leather onion rose Trial Odor Name 1 Odor Name 2 Odor Name 3 Odor Name 4 (romero) (cuero) (cebolla) (rosa) 4 strawberry (fresa) cedar wood (madera de cedro) chocolate (chocolate) baby powder (talco para bebes) 5 eggs (huevo) basil (albahaca) garlic (ajo) lilac (lila (flor)) 6 pineapple (pina) chlorine (cloro) sunblock (bloqueador solar) cotton candy (algodon de dulce) 7 chili pepper (aji picante) pine tree (pino) jasmine (jazmin) Smoke (humo) 8 new car smell (olor a carro nuevo) fresh bread (pan fresco) soap (jabon) mango (mango) 9 bam animal (animal de granja) wet paint (pintura humeda) grape (uva) apple (manzana) Note: Bold indicates the targeted and presented odor. Table 3: Odors and four choices presented to participants for the OPID18 test. Trial Odor Name 1 Odor Name 2 Odor Name 3 Odor Name 4 Practice trial cherry (cereza) glue (pegamento) grass (cesped) tomato sauce (salsa de tomate) 1 menthol (mentol) barbecue (parrilla) ginger (jengibre) dry-erase marker (marcador de borrado en seco) 2 new car smell (olor a carro nuevo) pine tree (pino) fresh bread (pan fresco) coffee mocha (cafe moca) 3 eggs (huevo) lavender (lavanda) raw fish (pescado cmdo) dirt (tierra) 4 rubbing alcohol (alcohol desinfectante) chocolate (chocolate) mango (mango) wet paint (pintura humeda) 5 honey (miel) lime (lima) soy sauce (salsa de soya) clove (clavo) 6 cheese (queso) cilantro (cilantro) onion (cebolla) orange (naranja) 7 cucumber (pepino) popcorn (palomitas de maiz) grape (uva) basil (albahaca) 8 bam animal (animal de granja) roast chicken (polio asado) cinnamon (canela) lilac (lila (flor)) 9 bubble gum (goma de mascar) fresh linen (lino fresco) garlic (ajo) cedar wood (madera de cedro) 10 cotton candy mulch (viruta / mantillo) black pepper (pimienta negra) leather (cuero) Trial Odor Name 1 Odor Name 2 Odor Name 3 Odor Name 4 (algodon de azucar) 11 ocean breeze (brisa del oceano) tobacco (tabaco) cola (coca cola) banana (banana) 12 vanilla (vainilla) jasmine (j azmin) smoke (humo) baby powder (talco para bebes) 13 soap (jabon) pumpkin (calabaza) coconut (coco) chlorine (cloro) 14 peppermint (menta) rose (rosa) watermelon (melon / sandia) chili pepper (aji picante) 15 licorice (regaliz (dulce)) bleach (blanqueador / cloro) pizza (pizza) lemon (limon) 16 olive (aceituna) caramel (caramelo) strawberry (fresa) sunblock (bloqueador solar) 17 apple (manzana) gasoline (gasolina) vinegar (vinagre) pineapple (pina) 18 peach (durazno / meloc oton) brown sugar (azucar morena) tar (brea / alquitran) rosemary (romero) Note: Bold indicates the targeted and presented odor. Example 2: Validation of unobserved remote testing of the AROMHA brain health test in cognitively normal individuals. In the first phase of self-administered testing, all participants completed the ABHT in an observed setting (remote via video conference (e.g., Zoom™) or in 5 person) during a scheduled appointment with the research assistant (n=70). Participants shared their testing screen and video with the research assistant via video conference during remote observed testing or completed the testing in person with their testing screen visible to the research assistant. In both these conditions, the research assistant observed and noted participant interactions with the software and 10 the cards and remained on standby if there were questions or confusion. Once confidence in the self-administered testing workflow was established, an unobserved (vs. observed) self-administration option was offered to interested, cognitively normal participants in order to validate the feasibility of this administration mode and the ability to scale future data collection. When cognitively healthy participants, enrolled 15 in the study, were given the option to self-administer remotely, independent of a research assistant, 70% chose to test on their own with the option of live help over the telephone, if needed. The overall distribution of participants by each administration modality is described in Table 4. Table 4: Distribution of Participants by Administration Modality. Anosmic (n=7) CN (n=127) see* (n=34) MCI (n=19) Total (n=187) Self-Administered 7 (100%) 127 (100%) 34(100%) 19(100%) 187 (100%) Remote 6 (86%) 118 (93%) 30 (88%) 18 (95%) 172 (92%) In-person 1 (14%) 9 (7%) 4(12%) 1 (5%) 15 (8%) Remote Observed 1 (14%) 62 (49%) 15 (44%) 16 (84%) 94 (50%) Remote Unobserved 5 (71%) 56 (44%) 15 (44%) 2(11%) 78 (42%) In person Observed 1 (14%) 9 (7%) 4(12%) 1 (5%) 15 (8%) * Three participants with SCC were removed from the main analyses because they were less than 55 years old. 5           When comparing olfactory scores between CN observed and unobserved groups, there were no significant differences in the olfactory outcomes after Bonferroni correction. In the unobserved modality group, age (p<0.001) was significantly greater and Spanish language ( / ?<0.001) was significantly less common, due to operational factors such as the availability of the unobserved option during 10 phases of recruitment, scheduling, and participant preference. Test duration ( / ?<0.001) was significantly greater in the unobserved group compared to the observed group. Sex and education were not different between the observed and unobserved groups (Table 5). Table 5: Demographic information and olfactory function in cognitively normal 15 participants who underwent observed and unobserved self-administration conditions of the AROMHA brain health test. CN Observed Self-administration (n=71) CN Unobserved Self-administration (n=56) P values Age (years) 44.89 (22.18) (19 to 93 years) 67.77(16.60) (20 to 92 years) <0.001 Sex (female %) 66% 64% 0.82 Education 16.80 (3.86) 16.68 (2.99) 0.84 Speaking language (Spanish %) 51% 9% <0.001 OPID9 ( / 9) 5.93 (1.38) 5.89 (1.40) 0.88 OPID9noguess ( / 9) 5.54 (1.58) 5.25 (1.92) 0.37 OPID18 ( / 18) 11.85 (2.79) 11.05 (2.57) 0.10 CN Observed Self-administration (n=71) CN Unobserved Self-administration (n=56) P values OPID18noguess ( / 18) 10.83 (3.24) 9.46 (3.57) 0.03 OD10 ( / 10) 8.59 (1.23) 8.36(1.41) 0.33 POEM (-1 to +1) 0.36 (0.32) 0.33 (0.28) 0.57 Average intensity ( / 10) 5.53 (1.83) 5.79(1.70) 0.41 Test Duration (minutes) 31.87 (8.70) 41.16(16.35) <0.001 Values are means (SD). T-tests were performed across groups for age, education, olfactory scores, and test duration. Chi-square test was performed on sex and speaking-language proportions across groups. As mean age significantly differed across observed and unobserved groups, olfactory scores were compared across groups, including age as a covariate. An ANCOVA revealed no significant interaction effects between age and groups for OPID18 (F [1, 123]=0.86, / >=0.36; partial r|2=0.01), OPID18noguess (F [1, 123]=2.16, 5      / ?=0.14; partial r| =0.02), OPID9 (F [1, 123]=0.41, / 1=0.52; partial r|2<0.001), OPID9noguess (F [1, 123]=2.64, / i=0.11; partial r|2=0.02), OD10 (F[l, 123]=1.92, / 1=0.17; partial r|2= 0.02), POEM (F [1, 123]=1.73, / i=0.19; partial r|2=0.01), or Intensity (F [1, 123]=1.33, / >=0.25; partial r|2=0.01). These results suggested that the effect of group (observed vs. unobserved) on olfactory scores was not moderated by 10 age. Example 3: Validation of the AROMHA brain health test in anosmic patients. Table 6 displays the demographics and olfactory scores of 7 patients with anosmia recruited from a smell loss clinic and CN participants. For context, the first column of Table 6 lists the score following random selection of the answers for each 15 test, e.g., the participant had no olfactory information to guide selection of the answers. Table 6: Distribution of olfactory scores of anosmic patients and CN participants on the AROMHA brain health test. Score of chance performance Anosmic (n=7) CN (n=12 7) p values Age (years) NA 54.14(20.28) (23 to 77 years) 54.98 (22.88) (19 to 93 years) 0.91 Sex (female %) NA 57% 65% 0.97 Education (years) NA 16.43 (2.44) 16.75 (3.49) 0.96 OPID9 ( / 9) 2.25 2.57 (0.79) 5.91 (1.38) <0.001 Score of chance performance Anosmic (n=7) CN (n=12 7) p values OPID9noguess ( / 9) 2.25 1.00(1.29) 5.41 (1.14) <0.001 OPID18 ( / 18) 4.5 5.00(1.29) 11.50(2.71) <0.001 OPID18noguess ( / 18) 4.5 1.14(1.46) 10.23 (3.45) <0.001 OD10 ( / 10) 5 6.14(1.57) 8.49(1.31) <0.001 POEM (-1 to +1) 0 -0.02 (0.10) 0.35 (0.30) <0.001 Average intensity ( / 10) NA 1.80 (1.33) 5.64(1.77) <0.001 Test Duration (min) NA 33.46 (11.33) 35.96 (13.42) 0.50 Values are means (SD). NA=not applicable. Wilcoxon signed-rank tests were performed across groups for age, education, olfactory scores, and test duration. The chi-square test was performed for sex. The anosmic group performed significantly worse on every olfactory metric ( / ?<0.001), as compared to the CN group although they did not take significantly longer to complete the battery when grouped across various modalities (Table 4). The anosmic group did not perform statistically differently from chance performance 5 on every olfactory measure, which was not the case for the CN control group. Example 4: Equivalence of the AROMHA brain health test in English vs. Spanish-speaking cognitively normal participants. When assessing the potential effect of language, no significant differences were found across olfactory scores between cognitively normal English-speaking and 10 Spanish-speaking participants. Demographically, only age was significantly different as the English-speaking group was significantly older than the Spanish-speaking group ( / 1=0.001) (Table 7). Table 7: Demographic information and olfactory function in cognitively normal English-speaking and Spanish-speaking participants. CN English-speaking participants (n=86) CN Spanishspeaking participants (n=41) p values Age (years) 59.16 (23.36) (19 to 93 years) 46.20 (19.32) (19 to 79 years) 0.001 Sex (female %) 63% 71% 0.38 Education 16.47 (2.84) 17.34 (4.56) 0.26 OPID9 ( / 9) 5.86 (1.46) 6.02 (1.21) 0.51 OPID9noguess ( / 9) 5.33 (1.86) 5.59 (1.47) 0.40 OPID18 ( / 18) 11.28 (2.92) 11.95 (2.18) 0.15 OPID18noguess ( / 18) 10.00 (3.62) 10.71 (3.04) 0.25 OD10 ( / 10) 8.42 (1.35) 8.63 (1.22) 0.37 CN English-speaking participants (n=86) CN Spanishspeaking participants (n=41) p values POEM (-1 to 1) 0.31 (0.30) 0.42 (0.30) 0.06 Average intensity ( / 10) 5.62 (1.77) 5.70 (1.80) 0.82 Test Duration (minutes) 36.28 (14.63) 35.30 (10.58) 0.67 Values are means (SD). T-tests were performed across groups for age, education, clinical variables, olfactory scores, and test duration. Chi-square test was performed on sex and speaking-language proportions across groups. As mean age significantly differed across language groups, olfactory scores were compared across groups, including age as a covariate. An ANCOVA revealed no significant interaction effects between age and language for OPID18 (F [1, 123]=1.54, / =0.22; partial p2 = 0.01), OPID18noguess (F [1, 123]=0.63, / =0.43; partial r|2 = 0.01), OPID9 (F [1, 123]=0.04, / =0.85; partial r|2 <0.001), OPID9noguess (F [1, 123 ]=1.25, / =0.27; partial p2 = 0.01), OD10 (F [1, 123]=0.73, / =0.39; partial r|2 = 0.01), POEM (F [1, 123]=3.42, / =007; partial p2 = 0.03), or Intensity (F [1, 123]=1.21, / 2=0.27; partial p2 = 0.01). These results suggested that the effect of language on olfactory scores was not moderated by age. Example 5: Diminished olfactory measured in the AROMHA brain health test with increasing age. When assessing the effect of age on olfaction, linear regression models of olfactory scores as a function of age in CN participants showed that greater age was significantly associated with lower OPID9noguess (fi = -0.02, / =0.002), OPID18 (fi = -0.04, / <0.001), lower OPID18noguess (fi = -0.06, / <0.001), lower OD10 ¢5 = -0.02, / <0.001), and lower POEM (fi = -0.004, / <0.001) scores, while the association with OPID9 (fi = -0.01, / =0.01) and average intensity (fi = -0.007, / =0.28) scores did not reach significance after Bonferroni correction (FIGS. 2A-2G). Example 6: Performance of AROMHA brain health test distinguished participants Aged +55 who were cognitively normal (CN), with subjective cognitive concerns (SCC) or with mild cognitive impairment (MCI). The sex and education of participants across the CN, SCC, and MCI groups were not significantly different after Bonferroni correction. Participants in the MCI group were older than those in the CN group ( / =0.01). Comparisons of each olfactory score of the ABHT for participants who were CN, had SCC, or had MCI revealed significant olfactory differences between subgroups, other than in the OD10, POEM odor memory score, and evaluations of average odor intensity (Table 8). Comparisons across cognitive groups in this +55 sample revealed no demographic 5 differences in sex and education, or in the test duration time. Table 8: Olfactory function across cognitive status among participants aged 55+. p values CN (n=73) SCC (n=31) MCI (n=19) CN vs. SCC CN vs. MCI SCC vs. MCI Age (years) (56 to 95 years) 1225 (9.13) 75.48 (9.22) 78.00 (8.07) 0.11 0.01 0.32 Sex (female %) 63% 62% 53% 0.99 0.57 0.76 Education (years) 16.42 (3.62) 16.26 (2.50) 16.53 (3.49) 0.78 0.91 0.77 OPID9 ( / 9) 5.71 (1.51) 5.19(1.45) 3.95 (1.43) 0.10 <0.001 0.005 OPID9noguess ( / 9) 5.16 (1.93) 4.29(1.79) 3.16 (1.77) 0.03 <0.001 0.04 OPID18 ( / 18) 10.79 (2.74) 10.68 (2.95) 7.26 (2.86) 0.62 <0.001 <0.001 OPID18noguess ( / 18) 9.4 (3.47) 8.03 (4.02) 5.16 (3.35) 0.11 <0.001 0.009 OD10 ( / 10) 8.12 (1.35) 7.42(1.52) 6.95 (1.96) 0.03 0.02 0.38 POEM (-1 to +nl) 0.27 (0.29) 0.24 (0.25) 0.22 (0.20) 0.61 0.37 0.72 Average intensity ( / 10) Test Duration (minutes) 5.55 (1.85) 40.10 (14.67) 4.93 (1.34) 40.49 (14.89) 4.71 (2.30) 40.69 (11.01) 0.06 0.66 0.15 0.85 0.71 0.83 Values are means (SD). T-tests were performed across groups for age, education, clinical variables, olfactory scores, and test duration. Chi-square test was performed on sex proportion across groups. Bold indicates statistically significant difference that remained significant after Bonferroni correction for 21 comparisons (p<0.002). When assessing the effect of cognitive status group on olfactory test components, ANCOVAs revealed a significant effect on OPID9, OPID9noguess, OPID18, OPID18noguess, and OD10 scores (Table 9). No effects of cognitive status 10 group were found for the POEM or average intensity scores. After Bonferroni correction for 21 comparisons ( / ?<0.002), post-hoc pairwise comparisons revealed significantly lower scores in the MCI group compared to CN older adults for the OPID9, OPID18, OPID18noguess, OD10 and to the SCC group for the OPID18 and OPID9 scores. Interaction effects between group and age were found for the OPID9noguess, 5 OPID18 no guess scores, and OD10; however, they did not remain significant after Bonferroni correction for 7 comparisons (p<0.007). No other significant interaction effects between group and age were found regarding other olfactory components (p>0.05). No other interaction effects with sex or education were found (p>0.05) (Table 9) 10 Table 9: ANCOVA models comparing different olfactory scores across groups. Source DF F P Post-hoc comparisons’ p values (Cohen’s d) CN vs SCC CN vs MCI SCC vs MCI OPID9 Group 2 11.07 <0.001 0.17 0.28 (1.09) < 0.001 (-4.44) 0.0017 (-3.29) Age 3.08 0.08 0.03 Sex 0.009 0.92 <0.001 Education 0.28 0.60 0.003 Group * Age 2 1.90 0.15 0.03 Group * Sex 2 0.22 0.80 0.004 Group * Education 2 2.37 0.10 0.04 Residuals 111 OPID9n oguess Group 2 10.35 <0.00 0.16 0.10 (1.68) < 0.001 (-4.20) 0.01 (-2.50) Age 1 8.92 0.003 0.07 Sex 1 0.80 0.37 0.007 Education 1 0.05 0.83 <0.001 Group * Age 2 3.96 0.02 0.07 Group * Sex 2 1.79 0.17 0.03 Group * Education 2 0.28 0.76 0.004 Residuals 111 OPID18 Group 2 12.84 <0.001 0.19 0.76 (0.30) < 0.001 (-4.91) < 0.001 (-4.19) Source DF F P Post-hoc comparisons’ p values (Cohen’s d) CN vs see CN vs MCI SCC vs MCI Age 1 2.49 0.12 0.02 Sex 1 2.53 0.11 0.02 Education 1 0.01 0.93 <0.001 Group * Age 2 3.34 0.04 0.06 Group * Sex 2 1.35 0.26 0.02 Group * Education 2 0.48 0.62 0.008 Residuals 111 OPID18n oguess Group 2 11.93 <0.001 0.18 0.16 (1.42) < 0.001 (-4.56) 0.003 (-3.29) Age 1 8.1 0.005 0.07 Sex 1 0.33 0.57 0.003 Education 1 0.18 0.68 0.002 Group * Age 2 4.79 0.01 0.08 Group * Sex 2 1.81 0.17 0.03 Group * Education 2 0.67 0.52 0.01 Residuals 111 OD10 Group 2 6.09 0.003 0.10 0.03 (2.19) 0.001 (-3.40) 0.14 (-1.50) Age 1 2.65 0.11 0.02 Sex 1 0.22 0.64 0.001 Education 1 0.001 0.96 <0.001 Group * Age 2 4.81 0.01 0.08 Group * Sex 2 1.08 0.34 0.02 Group * Education 2 1.10 0.34 0.02 Residuals 111 POEM Group 2 0.35 0.71 0.006 0.55 (0.55) 0.68 (-0.41) 0.96 (0.96) Age 1 3.44 0.07 0.03 Sex 1 0.44 0.51 0.004 Education 1 0.28 0.60 0.003 Group * Age 2 0.41 0.66 0.007 Group * Sex 2 2.83 0.06 0.05 Group * Education 2 0.52 0.60 0.009 Source DF F P Post-hoc comparisons’ / ? values (Cohen’s d) CN vs see CN vs MCI SCC vs MCI Residuals 111 Average Intensity Group 2 2.29 0.11 0.04 0.16 (1.40) 0.27 ( 1.09) 0.99 (0.003) Age 1 1.36 0.25 0.01 Sex 1 3.75 0.06 0.03 Education 1 0.007 0.93 <0.001 Group * Age 2 0.063 0.94 0.003 Group * Sex 2 0.052 0.95 <0.001 Group * Education 2 0.80 0.45 0.01 Residuals 111 DF=Degrees of Freedom, F=F-statistic, p=p-value, r|p2 = partial Eta squared. CI=Confidence interval. Bold represents post-hocp values that remained significant after Bonferroni correction for 21 comparisons (p<0.002). / > values from pairwise post-hoc comparisons adjusted for age, sex, and education. Example 7: Data driven machine learning approaches predicted cognitive impairment using many components of the AROMHA Brain Health Test. To control for age, sex, education, and comorbidities, we used both least 5 absolute shrinkage and selection operator (LASSO) and Gradient Boosting Machine (GBM) approaches to fit models on a training set (n = 78) and then predicted cognitive impairment in a held-out test set (n = 19). FIGS. 5A-5C show that, using a regularized logistic regression model, the AROMHA Brain Health Test predicted memory symptoms in subjects aged 20 - 95 (AUC = 0.94; FIGS. 5A and 5B) and in 10 subjects aged 55 -95 (AUC = 0.78; FIG. 5C). The GBM model iteratively constructed an ensemble of decision trees by fitting the error of the previous tree in the ensemble. The training process could be halted via an early stopping criterion to prevent overfitting to the training data. Hyperparameters of the GBM such as the total number of trees and the depth of each 15 tree were tuned via exhaustive grid search. GBMs, which can fit arbitrarily complex functional forms to accommodate non-linear relationships and interactions, predicted preclinical Alzheimer’s disease in subjects aged 55 -95 (AUC = 0.90). We then used a variable importance analysis to quantify the contribution of each input feature on the prediction of each test set example (FIG. 6). After age, OPID18 no guess provided the strongest signal in the variable importance analysis. These results indicated that the AROMHA Brain Health Test can predict cognitive impairment in preclinical Alzheimer’s disease (AD) and uses metrics not available by the University of Pennsylvania Smell Identification Test (UPSIT) or other conventional smell tests. Example 8: AROMHA Brain Health Assessment (ABHA) predicted computed brain volumes. Transactive Response DNA Binding Protein 43 kDa (TDP-43) pathology is frequently found in cases with Alzheimer’s disease (AD). TDP-43 pathology is associated with hippocampal and amygdala atrophy and greater AD severity. See, e.g., Meneses et al., MolNeurodegener. 2021 Dec 20; 16(1):84 and Huie et al., J Alzheimers Dis. 2023;91(4):1291-1301. As the amygdala volume is the first place of TDP-43 pathology in Alzheimer’s patients, use of the smell test to predict loss of amygdala volume can provide an early marker of TDP-43 pathology and, as such, a predictor of patient responsiveness to an AD treatment specific for the subset of AD patients with TDP-43 pathology. Of the participants that took the AROMHA Brain Health Test as described herein, 40 individuals (1) received an expertly ascertained clinical cognitive phenotype and (2) had MRI brain volume data available. In the present example, we present data on the n = 31 subjects that had a MRI scan date within + / - 3 years of the olfactory test. FIG. 7A shows the association of the OPID18 No Guess metric (score, x-axis) with the left hippocampus volume (y-axis). The left hippocampus discriminated the cognitive phenotype closest to the scan. This corroborated the findings that hippocampal volume was a powerful biomarker to detect early cognitive decline. The data suggest that because the AROMHA brain health test (smell test) could accurately predict hippocampal volume, then the smell test could also play a role in detecting early cognitive decline. FIG. 7B shows the association of the OPID18 No Guess metric (score, x-axis) with the left amygdala volume (y-axis). The data show that the AROMHA brain health test (smell test) could also predict amygdala volume according to cognitive phenotype. In FIG. 7B, we noted a cleaner gradation in density estimations for the cognitive phenotypes where normal (healthy) cognition had the highest mean and coverage of the top of the left amygdala volume range, subjective cognitive concern (SCC, or “impaired, not MCI (ImpNoMCI)” as shown on the graph) had the second highest mean, and mild cognitive impairment (MCI) had the lowest. The AROMHA OPID9 derived measure (the number out of 9 identification questions answered correctly when presented with forced choice options) was also found to be predictive for amygdala volume. Linear relationships between the OPID9 variable and the right amygdala volume are shown in FIG. 7C and linear relationships between the OPID9 variable and the left amygdala volume are shown in FIG. 7D. We noted stronger linear relationships between the OPID9 variable and the right amygdala volume relative to the OPID18 no guess variable. We also noted a uniformity of the slopes of the estimated linear fits for each of the cognitive phenotypes for both the right and left amygdala (FIGS. 7C-7D). Linear relationships between the OPID9 variable and the right hippocampal volume are shown in FIG. 7E and linear relationships between the OPID9 variable and the left hippocampal volume are shown in FIG. 7F. Age was a confounder as it was associated with poorer olfactory function and cognitive decline. The age distribution by cognitive phenotype closest to the scan date (FIG. 7G) showed clear distributional differences by cognitive label. Therefore, any prediction model leveraging the AROMHA Brain Health Test battery would need to demonstrate the olfactory metrics of the test have predictive utility via a pathway other than correlation with age. The strategy was thus to fit regularized linear regressions with a brain volume as the outcome to be predicted as a function of olfactory metrics derived from the AROMHA brain health test. If an olfactory metric was selected in a model fitting procedure, it was defined to have predictive utility beyond a correlation with age alone. The candidate predictor variables were OPID9 no guess, OPID9, OD10, OPID18.no.guess, OPID18.ID.score, total elapsed test time in minutes, OPID18.recall, preexisting condition index, age, average confidence, and sex. The volumes to be predicted were left hippocampus volume, right hippocampus volume, left amygdala volume, right amygdala volume, left thalamus volume, right thalamus volume, left entorhinal thickness, right enthorhinal thickness, right medialorbitofrontal thickness, and right medialorbitofrontal thickness. Each volume model was a 3-fold cross validated regularized linear regression fit using the glmnet library in R. (see, e.g., Friedman et al., 2010 Journal of Statistical Software, Articles 33 (1): 1-22; Simon et al., 2011 Journal of Statistical Software, Articles 39 (5): 1-13; and Tibshirani et al., 2012. Journal of the Royal Statistical Society: Series B (StatisticalMethodology) 74 (2): 245-66). Three (3)-fold cross validation divided the available data into 3 mutually exclusive equally sized subsets. Two (2) of the 3 subsets were used to train a linear model and the data from the third subset was used to test the model performance. Each fold played the role of the test set once. Performance was then averaged across the three folds. The cross validation procedure was used to select the hyperparameter lambda, the amount of penalization when fitting coefficients in the linear model. Heavy penalization implied a sparse solution where coefficients shrink to 0 and variables drop out of the model. No penalization amounted to a standard linear regression with no shrinkage in coefficients. The optimal lambda was the value that minimized the mean squared error average across the 3 folds. This value of lambda then implied a regularized linear model that could be used for future predictions. If an olfactory variable had a nonzero coefficient after the cross validation procedure, it improved the ability to predict the volumes on average. Cross validation could be sensitive to the random allocation of data across the folds. Therefore, the process could be repeated via sampling without replacement to quantify the stability of the selection process. We simulated this 500 times and plotted the fraction of times out of the 500 that each candidate variable was selected (had a nonzero coefficient at the optimal lambda) for each volume outcome. FIG. 7H shows the results from 500 simulations of cross validation procedure demonstrating that age, left amygdala, and OPID9 were selected almost always and thus carried the largest signal. These data demonstrated that for the left amygdala, age and OPID9 were almost always selected in the regularization process (FIG. 7H). An example of one specific model fit for left amygdala is shown in FIG. 71. In this particular regularized model as shown, only OPID9, age, and the model intercept were selected. Age had a negative coefficient, implying older individuals were predicted to have smaller left amygdala volumes. OPID9 had a positive coefficient, implying those that performed better on this AROMHA olfactory metric were predicted to have larger left amygdala volumes. For the left hippocampus, age and OPID9 were also almost always selected. The OPID18 recall and preexisting index were also selected in roughly half of the simulations. The right amygdala process selected age and OPID9 in most models. It also selected OPID10, OPID18 ID score, OPID18 recall, OPID9 No guess, sex, preexisting condition index, and total elapsed time in roughly half of all models. Collectively, these models implied that the AROMHA brain health test had improved predictive performance as measured via cross validation averaged mean squared error relative to models using age and sex alone. For scenarios where it would be meaningful to screen based on neuroanatomical volumes, the AROMHA brain health test could generate accurate predictions of volumes via a cost effective and accessible option. Example 9: Presence and immunogenicity of cdsRNA in AD with and without TDP-43 pathology. To assess the presence of cdsRNA in neurons of patients with AD pathology, we stained formalin-fixed paraffin-embedded (FFPE) amygdala sections of human postmortem brains of healthy individuals (n = 9) and of patients who met the pathologic diagnosis of AD (n = 10) (FIG. 8A). We stratified the patients and controls based on the severity of pTDP-43 pathology and detected cdsRNA in all individuals with severe pTDP-43 pathology by immunostaining. In contrast, whenever pTDP-43 inclusions were absent, our immunostainings were negative for cdsRNA as well (FIGS. 12A-12B). In individuals with mild pTDP-43 pathology, we detected cdsRNA in 33% of cases. Next, we evaluated the spatial coincidence of cdsRNA and pTDP-43 by performing cyclic immunofluorescence (CyCIF)34 which enabled multiplexed staining of pTDP-43 inclusions, cdsRNA, NFTs, A0 plaques, and nuclei on the same tissue section (FIG. 8B, lower panel of healthy control and AD, respectively). We compared the acquired images to luxol fast blue-hematoxylin and eosin (LHE) staining as well as immunohistochemical staining of cdsRNA, pTDP-43, tau, and A0 (FIG. 8B, upper panel of healthy control and AD, respectively). While NFTs and A0 plaques were detected in the tissue, neither pathology overlapped spatially with cytoplasmic pTDP-43 inclusions nor cdsRNA (FIG. 8B) However, pTDP-43 and cdsRNA were spatially coincident (enriched twofold in AD cases; FIG. 12C), paralleling our previous observations in ALS brains4. If this cdsRNA acted as a damage-associated molecular pattern (DAMP) in neurons, then it would trigger an IFN-I response through the activation of the PRRs RIG-I and MDA5 (FIG. 13A). To test whether cdsRNA was immunogenic in brains with AD pathology, we immunostained amygdala sections for phosphorylated protein kinase R (PKR), another PRR that binds to cdsRNA and activates autophosphorylation34 (FIG. 13B). We observed phosphorylated PKR only in cdsRNA-positive regions of AD brains, indicating a potential direct link between cdsRNA and the activation of PRRs. Furthermore, we wanted to see if we could replicate the observation of increased cdsRNA levels in a more controlled setting of TDP-43 pathology. For this purpose, we used the recently established human induced pluripotent stem cell (iPSC) line that harbored a pathogenic TDP-43 mutation (TDP-43+ / G298S)35 and also showed increased interferon signaling36. An isogenic pair of these iPSCs was differentiated into NGN2 cortical-like neurons36 and stained for cdsRNA (FIG. 8C). We observed a significant increase in dsRNA in cells with the familial TDP-43 mutation compared to control cells. As a positive control, we transfected the same control iPSC-derived neurons with poly(I:C), a dsRNA mimetic consisting of polyinosinic-polycytidylic acid, and saw a comparable increase in cdsRNA (FIG. 14). These observations identified shared characteristics of human AD-brain cells and two smaller, better controlled model systems for TDP-43 pathology. Next, we conducted a differential gene expression analysis of RNA-sequencing data from five relevant brain regions available from the Religious Order Study / Memory and Aging Project (ROSMAP)37 and Mount Sinai Brain Bank (MSBB)38 AMP-AD databases to test for elevated interferon signaling. We found 139 unique interferon-stimulated genes (ISGs) were significantly upregulated and 51 ISGs were significantly downregulated (adjusted P-value < 0.05) in AD patients compared to healthy controls. The parahippocampal gyrus (PHG), a brain region that shows atrophy early on in disease progression39,40, had the greatest number of upregulated ISGs, followed by the dorsolateral prefrontal cortex (DPFC) and the inferior frontal gyrus (IFG). In addition, we observed upregulation of ISGs, to a lesser extent, in the superior temporal gyrus (STG) and frontal pole (FP; FIGS. 8D and 15). Data demonstrated the presence of immunogenic cdsRNA-mediated IFN-I induction in AD-associated brain regions, although STING-based IFN-I induction may also be contributing to this signal41. Example 10: DRIAD-SP predicted efficacy of blocking interferon signaling. To examine whether interferon signaling in brains with AD pathology was associated with pTDP-43 pathology, we extended our DRIAD-SP framework along two axes. A machine learning framework for Drug Repurposing In Alzheimer’s Disease with Systems Pharmacology (DRIAD-SP)26 linked the prediction of disease stage (early- vs. late-stage AD based on the Braak AD staging system42) to drug-induced molecular signatures in human neural cells. Through this pipeline, we identified baricitinib, ruxolitinib, and tofacitinib, which were FDA-approved JAK inhibitors that inhibit interferon signaling, as drug candidates for repurposing in AD26 (FIG. 16A) We improved DRIAD-SP’s predictive power by training and evaluating predictors of Braak disease stage in a linear manner using ordinal ridge regression (FIG. 9A), instead of the binary classification, as was done previously. This approach allowed for training on the full scale of stage I to stage VI along the Braak spectrum42, without the need to stratify patients into “early” and “late” categories. Second, we used the expression of CEs as a proxy for the presence of pTDP-43 inclusions (FIG. 9B). Previous work demonstrated that the pre-mRNAs of stathmin-2 (STMN2)20 22 and UNCI3A18,19 have binding sites for TDP-43 and that TDP-43 represses CE inclusion in the canonical mRNA of both STMN2 and UNC13A during splicing. pTDP-43 pathology was associated with nuclear hypofunction of TDP-43 which led to CEs being included in the processed mRNA - resulting in a short variant of STMN2 mRNA20 22 (STMN2 short) due to a premature stop codon, and two versions of CE-inclusive UNC13 A mRNA, which we referred to as UNC13A-CE1 and UNC13A-CE218,19. Studies demonstrated the expression of CE-containing mRNA and de novo proteins also in AD27 32. Based on these findings, we posited the presence of these CEs in AD brains as a proxy for pTDP-43 pathology. We tested our computational approach by checking for the presence of CEs in RNA-sequencing data43 from FACS-sorted single neuronal nuclei from the neocortex of deceased ALS patients as about 97% of ALS patients exhibit pTDP-43 inclusions44. We confirmed that neuronal nuclei with loss of nuclear TDP-43 into cytoplasmic aggregates have higher reads of the CEs STMN2 short (log2 fold change (LFC) = 3.55; P < 0.0001), UNC13A-CE1 (LFC = 4.20; P = 0.0002) and UNC13A-CE2 (LFC = 4.05; P = 0.0006) relative to control neuronal nuclei that contain TDP-43 (FIG. 16B). By contrast, the number of reads of canonical STMN2 were reduced to a smaller degree (LFC = -1.76; P = 0.0017) and not reduced for UNC13A (LFC = 0.278; P = 0.11) mRNAs in TDP-43-deficient neuronal nuclei. With this validation, we stratified patients’ RNA-sequencing data derived from ROSMAP and MSBB AMP-AD databases based on the presence of CEs. First, we set the threshold for the presence or absence of each individual CE above the lowest non-zero peak (essentially > 1 CE transcript detected) in the abundance histograms (transcripts per million TPM >1.3 STMN2 short; TPM > 0.09 for UNC13A-CE1; TPM > 0.08 UNC13-CE2; FIG. 9C). We classified patients with two or more CE transcripts as TDP-43 positive (FIG. 9D). In the MSBB data, the counts for UNC13A-CE1 and UNC13A-CE2 were much lower relative to the ROSMAP data. We reasoned that the single-end sequencing protocol to generate the MSBB RNA-sequencing data was less sensitive for CE detection, relative to the paired-end reads in the ROSMAP data (FIG. 17). Consequently, we restricted our analysis to ROSMAP sequencing data from the posterior cingulate cortex (PCC), a brain region impaired early on in disease progression and therefore highly relevant for our study45. Of 546 total patients, no CEs were detectable in 85 patients, and one type of CE was detected in 257 patients; 183 patients had two types of CEs, and 21 patients had all three CEs present (FIG. 9D). Hence, 37% of cases were classified “TDP-43-positive”-AD patients, which is a conservative prediction considering that the clinical incidence of TDP-43-pathology in AD patients can reach 57%14. To determine whether predicted TDP-43 positivity was associated with later Braak stages, we used an ordinal regression model and found no significant difference in Braak staging (P = 0.27) between “positive” and “negative” cases (FIG. 18A), indicating that CE production was orthogonal to the stage of tau progression. We assessed the predicted efficacy of baricitinib, ruxolitinib and tofacitinib in AD patients with 2 or more CEs (as a proxy of TDP-43 dysfunction) relative to AD patients with one or no CEs (FIG. 9E, further replicates in FIG. 18B). Both baricitinib and ruxolitinib, but not tofacitinib, were significant hits in the predictor trained on RNA-sequencing data from AD patients with 2 or more CEs. On the other hand, none of the three drugs (baricitinib P = 0.76, ruxolitinib P = 0.34, tofacitinib P = 0.16) were significant hits in the predictor trained on no / low CE-expressing AD patients. These findings support that cdsRNA associated with TDP-43 dysfunction contributed to IFN-I-mediated neuroinflammatory processes occurring in AD patients’ brains. Example 11: CRISPR screen and validation of TYK2. To replicate cdsRNA-mediated inflammation and further determine that cdsRNA was not only immunogenic but also triggered neural cell death, we transfected differentiated ReN VM cells with poly(I:C). These differentiated human neural cells contained markers for neurons, astrocytes, and oligodendrocytes, but not microglia46 and were not sensitive to other common stressors such as DNA-damaging agents and reactive oxygen species (FIGS. 19A-19J), perhaps because this cell line does not express STING46. In contrast to our previous work (“One Pot” Differentiation, FIG. 20A)4, we found that differentiating neural cells in a separate dish and then transferring differentiated cells into the assay plate reduced variability between technical replicates (“Separate Pot” Differentiation; FIG. 20B). Both baricitinib and ruxolitinib rescued human neural cell death in this new workflow with greater reproducibility and lower variance with nearly complete rescue at 1 pM and > 100% at 10 pM - presumably due to preventing toxicity from the transfection reagent lipofectamine 2000 present in the control. Both drugs significantly inhibited the phosphorylation of STAT1 and the downstream transcription of ISGs (FIGS. 21A-21D) Since the ECso values of baricitinib (ECso = 199.6 nM) and ruxolitinib (ECso = 360.7 nM) were much greater than their affinities to their target proteins (baricitinib: IC5o(JAK1) = 5.9 nM, IC5o(JAK2) = 5.7 nM47; ruxolitinib: IC5o(JAK1) = 3.3 nM, ICso(JAK2) = 2.8 nM48), the inhibition of other kinases may be responsible for their neuroprotective effect. To test this, we conducted a genome-wide CRISPR screen to rescue cdsRNA-induced toxicity in differentiated human neural cells4,49. We used the Brunello library which consisted of four single guide RNAs (sgRNAs) per human gene and 1,000 control sgRNAs50 (FIG. 10A) expressed in barcoded lentiviral particles to transduce ReN VM cells in their neural progenitor state. After a weeklong differentiation, one replicate was harvested to identify genes that were essential for ReN cell differentiation and were not present at the initiation of either poly(I:C) or lipofectamine. Cells carrying sgRNA targeting eight different genes were significantly depleted post-differentiation (DGAT2, SIGLECU, GSTA2, ABH, FBXL8, AHCY, C7orf55-LUC7L2, ARHGEF12), indicating that these genes may be essential for neural cell survival and differentiation. These post-differentiation cells were treated for 48 hours with poly(I:C) or mock lipofectamine transfection, were harvested, genomic DNA was extracted, and the sgRNA sequences within the gDNA were PCR-amplified and sequenced to identify which sgRNA barcodes are enriched. Enriched sgRNAs in the post-poly(I:C) treatment arm versus the mock transfection arm correspond to candidate target genes whose loss-of-function enhances survival. We found interferon receptor alpha and beta subunit 2 (IFNAR2\ average LFC = 2.23825, P = 0.0001), interferon regulatory factor 9 (IRF9', average LFC = 2.0625, P = 0.0002), and TYK2 (average LFC = 1.95, P = 0.0004) were the best candidates for neural cell rescue upon knockout, whereas knockout of phospholipase C gamma 1 (PLCGP, average LFC = -1.9875, P = 0.0002), hepatocyte growth factor-regulated tyrosine kinase substrate (HGS; average LFC = -1.775, P = 0.0004), and SQSTM1 (p62, average LFC = -1.7475, P = 0.0005) markedly increased the toxicity induced by poly(I:C) (FIG. 10B). We independently validated IFNAR2 and TYK2 using complementary chemical probes. To evaluate IFNAR2, we treated ReN VM cell-derived neural cells with an anti-IFNAR2 antibody before introducing poly(I:C) into the cells (FIG. 10C). The blocking antibody significantly rescued dsRNA-mediated neuronal death, but the response was not dose-dependent and plateaued at about 50% rescue (no antibody P < 0.0001; antibody 1:50 P = 0.0001; 1:25 P = 0.0002; 1:10 P = 0.0011). This was consistent with our observation that activating IFNAR2 with interferon-alpha was not sufficient to trigger human neural cell death. We treated cells with interferon-a at a range of doses. After 48 hours, there was no decrease in cell viability at any dose tested (FIG. 21E). To evaluate TYK2, we used deucravacitinib, a potent small molecule inhibitor 33 that was highly specific for TYK2 over other JAK kinase family members (IC50 = 0.2 nM compared to IC5o(JAK1) = 1 nM and IC5o(JAK2) > 200 nM)51’52. We observed that exposure to deucravacitinib rescued dsRNA-toxicity up to 100% at all tested concentrations (FIG. 10D)with an EC50 <10 nM. Deucravacitinib was also found to inhibit STAT1 phosphorylation (downstream protein of TYK2) completely. (FIG. 10E, all P < 0.0001). We validated these results in differentiated ReN CX cells, a cell line that showed the same characteristics as ReN VM cells but was derived from the cortex instead of the ventral mesencephalon and generated glutamatergic neurons rather than dopaminergic neurons, and in SH-SY5Y cells, a neuroblastoma cell line (FIGS. 21F-21I) In both cell types, full rescue of cdsRNA-mediated toxicity was achieved with a 1 pM dose of deucravacitinib. Higher concentrations of deucravacitinib seemed to be toxic to these cells. To further evaluate TYK2 as a key signaling kinase for IFN-I signaling, we performed a deep proteomics analysis from ReN VM cell-derived neural cells that were treated with baricitinib and deucravacitinib at high doses (10 pM) by quantitative multiplex tandem mass tag (TMT) mass spectrometry (FIG. 10F). Here, we saw that all quantified ISGs were upregulated by poly(I:C), and this ISG induction was reversed by both baricitinib and deucravacitinib. Taken together, our results indicated that the toxic innate immune reaction to cdsRNA required the activation of IFNAR and was dependent on TYK2. Furthermore, the action of TYK2 appeared to be more complex than just mediating the IFN-I pathway alone. In sum, the CRISPR screen analysis provided herein implicated the selective inhibition of TYK2 as a target of rescuing cdsRNA-induced toxicity in cdsRNA / TDP-43-positive neurodegenerative diseases (e.g., AD patients with TDP-43 pathology). Example 12: Biomarkers for dsRNA pathology. Specific biomarkers for neuroinflammatory-based neurodegenerative disease subtypes associated with dsRNA-induced pathology would help identify patients who could benefit from treatments modulating TYK2. We screened potential biomarkers in a genome-wide association study (GWAS) using proteomics data from the UK Biobank53 and Icelandic DeCODE database54. Even though processed on different proteomics platforms (UK Biobank: Olink, DeCODE: SomaScan), we consistently found that individuals with an established partial loss-of-function polymorphism in TYK2 (rs34536443) in both cohorts have significantly lower plasma CXCL10 levels (UK Biobank: Effect: -0.12, SE = 0.02, P = 2.72 x 10’11; DeCODE: Effect: -0.10, SE = 0.02, P = 5.04 x 10'11) (FIG. 11A). This minor allele coding mutation (Pl 104A) led to a near-complete (-80%) loss of TYK2 function in homozygotes, while heterozygotes have <40% reduction of function55 57. We also examined potential biomarkers in cell-based assays by conducting Meso Scale Discovery (MSD) immunoassays on a selected array of ISG products. We quantified CXCL10, CCL2, and IL-6 in media taken from ReN VM cells transfected with poly(I:C) and treated with baricitinib, ruxolitinib, deucravacitinib, or vehicle (FIGS. 11B-11D) The dsRNAmi increased the level of these candidate biomarkers, and the drugs diminished their expression (P < 0.0001 in all conditions) to levels comparable to the lipofectamine-only control condition. Next, to evaluate if this candidate biomarker can be seen in association with TDP-43 pathology, we modified our cell-based assay by treating differentiated ReN VM cells with 1 pM MG-132, a proteasome inhibitor known to induce cytoplasmic aggregation of TDP-4361,62. We confirmed translocation of TDP-43 from the nucleus to the cytoplasm by immunofluorescent staining (FIG. 22) and quantified CXCL10 levels in the media by ECL immunoassay. CXCL10 levels were significantly elevated after MG-132 treatment (P < 0.005) and were reduced to control levels by 10 pM deucravacitinib (P < 0.005) but not by baricitinib and ruxolitinib treatment, indicating that selective TYK2 inhibition was most effective in reducing IFN-I mediated CXCL10 levels among the JAK kinases (FIG. HE). Taken together, these results nominated CXCL10, CCL2, and IL-6 as candidate biomarkers for the assessment of dsRNA-mediated neuroinflammation using the CSF of patients with AD and could help determine whether these patients would benefit from therapeutic approaches reducing TYK2 activity (FIG. 11F). Example 13: The AROMHA olfactory battery was a strong discriminator of clinical cognitive phenotype relative to the biomarker, ptau217. Of the participants that have taken the AROMHA brain health test as described herein, 51 individuals had received (1) a clinical cognitive phenotype ascertained by expert clinical investigators and affirmed in a consensus diagnostic conference at the Massachusetts Alzheimer’s Disease Research Center and (2) plasma biomarker quantification for phospho-tau217 (ptau217), a biomarker of interest for early detection of Alzheimer’s Disease pathology. Presented herein are data on the n = 27 individuals that had a biomarker draw within + / -3 years of the olfactory test date. This time interval was chosen as it was hypothesized that the rate of change of biomarkers would not vary substantially relative to the timing of the olfactory testing. FIG. 23 demonstrates the association of the derived ‘OPID18 No Guess’ variable (defined as the number of olfactory identification questions answered correctly out of 18 possible questions without attestation of guessing amongst the forced choice options) with ptau217. FIG. 23 also displays the marginal density estimation of each respective variable along the axes colored by the expertly ascertained cognitive label closest in time to the biomarker draw. 24 out of 27 individuals had coincident biomarker draws with cognitive label ascertainment. The remaining 3 had a cognitive ascertainment within 0.6 years on either side of the biomarker draw. From these data, it was noted that the OPID18 No Guess variable was able to discriminate between the 3 cognitive phenotype classes better than ptau217. REFERENCES 1.     2023 Alzheimer’s disease facts and figures. Alzheimer’s & Dementia 19, 1598-1695 (2023). 2. Jack, C. R. et al. Longitudinal tau PET in ageing and Alzheimer’s disease. Brain 141, 1517-1528 (2018). 3. Bateman, R. J. et al. Clinical and Biomarker Changes in Dominantly Inherited Alzheimer’s Disease, n engl j med 10 (2012). 4. Villemagne, V. L. et al. Amyloid P deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: a prospective cohort study. 12, 11 (2013). 5. Jia, J. et al. Biomarker Changes during 20 Years Preceding Alzheimer’s Disease. N Engl J Med 390, 712-722 (2024). 6. Munro, C. E. et al. Recent contributions to the field of subjective cognitive decline in aging: A literature review. Alz & Dem DiagAss & Dis Mo 15, e 12475 (2023). 7. Knopman, D. S. et al. Alzheimer disease. Nat Rev Dis Primers 7, 33 (2021). 8. Cummings, J., Feldman, H. H. & Scheltens, P. The “rights” of precision drug development for Alzheimer’s disease. Alz Res Therapy 11, 76 (2019). 9. Sperling, R. A., Jack, C. R. & Aisen, P. S. Testing the Right Target and Right Drug at the Right Stage. Sci. Transl. Med. 3, (2011). 10. Kim, C. K. et al. Alzheimer’s disease: key insights from two decades of clinical trial failures. Journal of Alzheimer’s Disease 87, 83-100 (2022). 11. Cummings, J. et al. Re-engineering Alzheimer clinical trials: Global Alzheimer’s Platform network. The journal of prevention of Alzheimer’s disease 3, 114 (2016). 12. Reiss, A. B. et al. Alzheimer’s disease: many failed trials, so where do we go from here? Journal of Investigative medicine 68, 1135-1140 (2020). 13. Murphy, C. Olfactory and other sensory impairments in Alzheimer disease. Nat Rev NeuroIIS, 11-24 (2019). 14. Albers, M. W., Tabert, M. H. & Devanand, D. P. Olfactory dysfunction as a predictor of neurodegenerative disease. Curr Neurol Neurosci Rep 6, 379-386 (2006). 15. Pashkovski, S. L. et al. Structure and flexibility in cortical representations of odour space. Nature 583, 253-258 (2020). 16. Sosulski, D. L., Bloom, M. L., Cutforth, T., Axel, R. & Datta, S. R. Distinct representations of olfactory information in different cortical centres. Nature 472, 213-216 (2011). 17. Braak, H. & Braak, E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 82, 239-259 (1991). 18. Diez, I. et al. Tau propagation in the brain olfactory circuits is associated with smell perception changes in aging. Nat Commun 15, 4809 (2024). 19. Arnold, S. E. et al. Olfactory epithelium amyloid-P and paired helical fdament-tau pathology in Alzheimer disease. Ann Neurol. 67, 462-469 (2010). 20. Jobin, B., Boiler, B. & Frasnelli, J. Volumetry of Olfactory Structures in Mild Cognitive Impairment and Alzheimer’s Disease: A Systematic Review and a Meta-Analysis. Brain Sciences 11, 1010 (2021). 21.     Jobin, B., Boiler, B. & Frasnelli, J. Smaller grey matter volume in the central olfactory system in mild cognitive impairment. Experimental Gerontology (2023). 22. Lu et al. Functional Connectivity between the Resting-State Olfactory Network and the Hippocampus in Alzheimer’s Disease. Brain Sciences 9, 338 (2019). 23. Vasavada, M. M. et al. Olfactory cortex degeneration in Alzheimer’s disease and mild cognitive impairment. Journal of Alzheimer’s disease 45, 947-958 (2015). 24. Doty, R. L., Shaman, P., Kimmehnan, C. P. & Dann, M. S. University of Pennsylvania Smell Identification Test: a rapid quantitative olfactory function test for the clinic. The Laryngoscope 94, 176-178 (1984). 25. Hummel, T., Sekinger, B., Wolf, S. R., Pauli, E. & Kobal, G. ‘Sniffin’sticks’: olfactory performance assessed by the combined testing of odor identification, odor discrimination and olfactory threshold. Chemical senses 22, 39-52 (1997). 26. Lafaille-Magnan, M.-E. et al. Odor identification as a biomarker of preclinical AD in older adults at risk. Neurology 89, 327-335 (2017). 27. Risacher, S. L. et al. Olfactory identification in subjective cognitive decline and mild cognitive impairment: Association with tau but not amyloid positron emission tomography. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring 9, 57-66 (2017). 28. Tu, L. et al. Association of Odor Identification Ability With Amyloid-P and Tau Burden: A Systematic Review and Meta-Analysis. Front. Neurosci. 14, 586330 (2020). 29. Klein, J. et al. Olfactory Impairment Is Related to Tau Pathology and Neuroinflammation in Alzheimer’s Disease. JAD 80, 1051-1065 (2021). 30. Reijs, B. L. R. et al. Relation of Odor Identification with Alzheimer’s Disease Markers in Cerebrospinal Fluid and Cognition. JAD 60, 1025-1034 (2017). 31. Devanand, D. P. et al. Olfactory identification deficits and MCI in a multi-ethnic elderly community sample. Neurohiology of Aging 31, 1593-1600 (2010). 32. Kose, Y. et al. Association between the inability to identify particular odors and physical performance, cognitive function, and / or brain atrophy in community-dwelling older adults from the Fukuoka Island City study. BMC Geriatr 21, 421 (2021). 33. Hagemeier, J. et al. Odor identification deficit in mild cognitive impairment and Alzheimer’s disease is associated with hippocampal and deep gray matter atrophy. Psychiatry Research: Neuroimaging 255, 87-93 (2016). 34. Kjelvik, G. et al. The brain structural and cognitive basis of odor identification deficits in mild cognitive impairment and Alzheimer’s disease. BMC Neurol 14, 168 (2014). 35. Murphy, C., Jernigan, T. L. & Fennema-Notestine, C. Left hippocampal volume loss in Alzheimer’s disease is reflected in performance on odor identification: A structural MRI study. JInt Neuropsychol Soc 9, 459-471 (2003). 36. Yoshii, F., Onaka, H., Kohara, S., Ryo, M. & Takahashi, W. Association of Smell Identification Deficit with Alzheimer’s Disease Assessment Scale-Cognitive Subscale, Japanese Version Scores and Brain Atrophy in Patients with Dementia. Eur Neurol 81, 145151 (2019). 37. Yu, H., Chen, Z., Zhao, J., Duan, S. & Zhao, J. Olfactory Impairment and Hippocampal Volume in a Chinese MCI Clinical Sample. Alzheimer Disease & Associated Disorders 33, 124-128 (2019). 38. Jobin, B., Roy-C6te, F., Frasnelli, J. & Boiler, B. Olfaction and Declarative Memory in Aging: A Meta-analysis. Chemical Senses bjad045 (2023) doi: 10.1093 / chemse / bjad045. 39. Tabert, M. H. et al. A 10-item smell identification scale related to risk for Alzheimer’s disease. Ann Neurol. 58, 155-160 (2005). 40. Devanand, D. P. et al. Combining Early Markers Strongly Predicts Conversion from Mild Cognitive Impairment to Alzheimer’s Disease. Biological Psychiatry 64, 871-879 (2008). 41. Conti, M. Z. et al. Odor Identification Deficit Predicts Clinical Conversion from Mild Cognitive Impairment to Dementia Due to Alzheimer’s Disease. Archives of Clinical Neuropsychology 28, 391-399 (2013). 42. Roberts, R. O. et al. Association between olfactory dysfunction and amnestic mild cognitive impairment and Alzheimer disease dementia. JAMA neurology 73, 93-101 (2016). 43. Dhilla Albers, A. et al. Episodic memory of odors stratifies Alzheimer biomarkers in normal elderly: POEM: Odor Memory Biomarker in Normal Elderly. Ann Neurol. 80, 846857(2016). 44. Devanand, D. P. et al. Olfactory deficits predict cognitive decline and Alzheimer dementia in an urban community. Neurology 84, 182-189 (2015). 45. Dintica, C. S. et al. Impaired olfaction is associated with cognitive decline and neurodegeneration in the brain. Neurology 92, e700-e709 (2019). 46. Growdon, M. E. et al. Odor identification and Alzheimer disease biomarkers in clinically normal elderly. Neurology 84, 2153-2160 (2015). 47. Olofsson, J. K., Larsson, M., Roa, C., Wilson, D. A. & Jonsson Laukka, E. Interaction Between Odor Identification Deficit and APOE4 Predicts 6-Year Cognitive Decline in Elderly Individuals. Behav Genet 50, 3-13 (2020). 48. Sohrabi, H. R. et al. Olfactory discrimination predicts cognitive decline among community-dwelling older adults. Transl Psychiatry 2, e 118—e 118 (2012). 49. Windon, M. J., Kim, S. J., Oh, E. S. & Lin, S. Y. Predictive value of olfactory impairment for cognitive decline among cognitively normal adults. The Laryngoscope 130, 840-847 (2019). 50. Wheeler, P. L. & Murphy, C. Olfactory Measures as Predictors of Conversion to Mild Cognitive Impairment and Alzheimer’s Disease. Brain Sciences 11, 1391 (2021). 51. Wilson, R.S.et al. Olfactory Identification and Incidence of Mild Cognitive Impairment in Older Age. Arch Gen Psychiatry 64, 802 (2007). 52. Rahayel, S., Frasnelli, J. & Joubert, S. The effect of Alzheimer’s disease and Parkinson’s disease on olfaction: A meta-analysis. Behavioural Brain Research 231, 60-74 (2012). 53. Schofield, P. W., Moore, T. M. & Gardner, A. Traumatic Brain Injury and Olfaction: A Systematic Review. Front. Neurol. 5, (2014). 54. Audronyte, E., Pakulaite-Kazliene, G., Sutnikiene, V. & Kaubrys, G. Odor Discrimination as a Marker of Early Alzheimer’s Disease. JAD 1-10 (2023) doi:10.3233 / JAD-230077. 55. Bastin, C. et al. Anosognosia in Mild Cognitive Impairment: Lack of Awareness of Memory Difficulties Characterizes Prodromal Alzheimer’s Disease. Front. Psychiatry 12, 631518 (2021). 56. Gallo, D. A., Cramer, S. J., Wong, J. T. & Bennett, D. A. Alzheimer’s disease can spare local metacognition despite global anosognosia: Revisiting the confidence-accuracy relationship in episodic memory. Neuropsychologia 50, 2356-2364 (2012). 57. Cosentino, S., Metcalfe, J., Butterfield, B. & Stem, Y. Objective Metamemory Testing Captures Awareness of Deficit in Alzheimer’s Disease. Cortex 43, 1004-1019 (2007). 58. Vannini, P. et al. Decreased meta-memory is associated with early tauopathy in cognitively unimpaired older adults. NeuroImage: Clinical 24, 102097 (2019). 59. Lopez-Martos, D. et al. Awareness of episodic memory and meta-cognitive profiles: associations with cerebrospinal fluid biomarkers at the preclinical stage of the Alzheimer’s continuum. Front. Aging Neurosci. 16, 1394460 (2024). 60. Albers, M. W. et al. At the interface of sensory and motor dysfunctions and Alzheimer’s disease. Alzheimer’s & Dementia 11, 70-98 (2015). 61. Patel, Z. M. et al. International consensus statement on allergy and rhinology: Olfaction. Int Forum Allergy Rhinol 12, 327-680 (2022). 62. Rodriguez, S. et al. Innate immune signaling in the olfactory epithelium reduces odorant receptor levels: modeling transient smell loss in COVID-19 patients. Preprint at https: / / doi.org / 10.1101 / 2020.06.14.20131128 (2020). 63. Ergun, B. et al. Longitudinal Assessment Of Subjective And Objective Changes In Olfactory Function Following Sars-Cov-2 Infection: A Focus On Odor Identification, Intensity, And General Smell Function, in (Bonita Springs, FL, 2024). 64. Jobin, B. et al. Olfactory function is predictive of brain volumes and memory of former professional football players in the Harvard Football Players Health Study, in (Reykjavik, Iceland., 2024). 65. Runde, A. et al. Digital Accessible Remote Olfactory Mediated Health Assessments For Preclinical AD. in (Bonita Springs, FL, 2024). 66. Jessen, F. et al. A conceptual framework for research on subjective cognitive decline in preclinical Alzheimer’s disease. Alzheimer’s & Dementia 10, 844-852 (2014). 67. Sperling, R. A. et al. Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & Dementia 7, 280-292 (2011). 68.    Jonsson, F. U. Olfactory Metacognition. Chemical Senses 28, 651-658 (2003). 69.    Jonsson, F. U. Odor Emotionality Affects the Confidence in Odor Naming. Chemical Senses 30, 29-35 (2005). 70. Roalf, D. R. et al. A quantitative meta-analysis of olfactory dysfunction in mild cognitive impairment. J Neurol Neurosurg Psychiatry 88, 226-232 (2017). 71. Kjelvik, G. et al. The Human Brain Representation of Odor Identification in Amnestic Mild Cognitive Impairment and Alzheimer’s Dementia of Mild Degree. Frontiers in Neurology 11, 1779 (2021). 72. Patin, A. & Pause, B. M. Human amygdala activations during nasal chemoreception. Neuropsychologia 78, 171-194 (2015). 73. Larsson, M. et al. Olfactory memory in the old and very old: relations to episodic and semantic memory and APOE genotype. Neurohiology of Aging 38, 118-126 (2016). 74. Martin, C., Beshel, J. & Kay, L. M. An Olfacto-Hippocampal Network Is Dynamically Involved in Odor-Discrimination Learning. Journal of Neurophysiology 98, 2196-2205 (2007). 75. Desiato, V. M. et al. The Prevalence of Olfactory Dysfunction in the General Population: A Systematic Review and Meta-analysis. Am JRhinol Allergy 35, 195-205 (2021). 76. Zhang, C. & Wang, X. Initiation of the age-related decline of odor identification in humans: A meta-analysis. Ageing Research Reviews 40, 45-50 (2017). 77. Oleszkiewicz, A., Schriever, V., Croy, 1., Hahner, A. & Hummel, T. Updated Sniffin’Sticks normative data based on an extended sample of 9139 subjects. European Archives of Oto-Rhino-Laryngology 276, 719-728 (2019). 78. Moberg, P. J. & Raz, N. Aging and Olfactory Recognition Memory: Effect of Encoding Strategies and Cognitive Abilities. International Journal of Neuroscience 90, 277291 (1997). 79. Murphy, C., Nordin, S. & Acosta, L. Odor learning, recall, and recognition memory in young and elderly adults. Neuropsychology 11, 126 (1997). 80. Rovee, C. K., Cohen, R. Y. & Schlapack, W. Life-span stability in olfactory sensitivity. Developmental Psychology 11, 311 (1975). 81. Cowart, B. J. Relationships between taste and smell across the adult life span a. Annals of the New York academy of Sciences 561, 39-55 (1989). 82. Murphy, C. Age-related effects on the threshold, psychophysical function, and pleasantness of menthol. Journal of Gerontology 38, 217-222 (1983). 83. Frasnelli, J. et al. Neuroanatomical correlates of olfactory performance. Exp Brain Res 201, 1-11 (2010). 84. Savic, I., Gulyas, B., Larsson, M. & Roland, P. Olfactory Functions Are Mediated by Parallel and Hierarchical Processing. Neuron 26, 735-745 (2000). 85. Wilson, D. A. et al. Cortical Odor Processing in Health and Disease, in Progress in Brain Research vol. 208 275-305 (Elsevier, 2014). 86. Haehner, A., Hummel, T. & Reichmann, H. Olfactory Loss in Parkinson's Disease. Parkinson’s Disease 2011, 450939 (2011). 87. Yoo, H. S. et al. Olfactory dysfunction in Alzheimer’s disease- and Lewy body-related cognitive impairment. Alzheimer’s & Dementia 14, 1243-1252 (2018). 88. Camemolla, S. E. et al. Olfactory dysfunction in frontotemporal dementia and psychiatric disorders: A systematic review. Neuroscience & Biohehavioral Reviews 118, 588— 611 (2020). 89. Lecuyer Giguere, F. et al. Early Parosmia Signs and Affective States Predict Depression and Anxiety Symptoms 6 Months After a Mild Traumatic Brain Injury. Chemical Senses 45, 483-490 (2020). 90. Alosco, M. L. et al. Olfactory Function and Associated Clinical Correlates in Former National Football League Players. Journal of Neurotrauma 34, 772-780 (2017). 91. Zigrand, C. et al. Olfactory perception in patients with a mild traumatic brain injury: a longitudinal study. Brain Injury 36, 985-990 (2022). 92. Doraiswamy, P. M., Narayan, V. A. & Manji, H. K. Mobile and pervasive computing technologies and the future of Alzheimer’s clinical trials. Npj Digital Medicine 1, 1 (2018). 93. Kaye, J. et al. Using digital tools to advance Alzheimer’s drug trials during a pandemic: the EU / US CTAD task force. The journal ofprevention of Alzheimer’s disease 8, 513-519 (2021). 94. Green, P., Rohling, M. L., Iverson, G. L. & Gervais, R. O. Relationships between olfactory discrimination and head injury severity. Brain Injury 17, 479-496 (2003). 95. Tsai, M. S. et al. Apolipoprotein E: risk factor for Alzheimer disease. Am J Hum Genet 54, 643-649 (1994). 96. Weiner, M. W. et al. Increasing participant diversity in AD research: plans for digital screening, blood testing, and a community-engaged approach in the Alzheimer’s Disease Neuroimaging Initiative 4. Alzheimer’s & Dementia 19, 307-317 (2023). 97. Kruse, C. S. et al. Telehealth and patient satisfaction: a systematic review and narrative analysis. BMJ Open 7, eO 16242 (2017). 98. Atmojo, J. T. et al. Telemedicine, Cost Effectiveness, and Patients Satisfaction: A Systematic Review. J HEALTH POLICY MANAGE 5, 103-107 (2020). 99. Besser, L. et al. Version 3 of the National Alzheimer’s Coordinating Center’s Uniform Data Set. Alzheimer Disease & Associated Disorders 32, 351-358 (2018). 100. Albert, M. S. et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimer’s & Dementia 7, 270-279 (2011). 1. Chen, X. & Holtzman, D. M. Emerging roles of innate and adaptive immunity in Alzheimer’s disease. Immunity 55, 2236-2254 (2022). 2. Hansen, D. V., Hanson, J. E. & Sheng, M. Microglia in Alzheimer’s disease. J Cell Biol 217, 459(2018). 3. Mead, R. J., Shan, N., Reiser, H. J., Marshall, F. & Shaw, P. J. Amyotrophic lateral sclerosis: a neurodegenerative disorder poised for successful therapeutic translation. Nature Reviews Drug Discovery 2022 22:3 22, 185-212 (2022). 4. Rodriguez, S. et al. Genome-encoded cytoplasmic double-stranded RNAs, found in C9ORF72 ALS-FTD brain, propagate neuronal loss. Sci Transl Med 13, (2021). 5. Palmer, A. C. & Sorger, P. K. Combination Cancer Therapy Can Confer Benefit via Patient-to-Patient Variability without Drug Additivity or Synergy. Cell 171, 1678-1691.el3 (2017). 6. Moscow, J. A., Fojo, T. & Schilsky, R. L. The evidence framework for precision cancer medicine. Nature Reviews Clinical Oncology 2017 15:3 15, 183-192 (2017). 7. Tsimberidou, A. M., Fountzilas, E., Nikanjam, M. & Kurzrock, R. Review of precision cancer medicine: Evolution of the treatment paradigm. Cancer Treat Rev 86, (2020). 8. Andrews, S. J. et al. The complex genetic architecture of Alzheimer’s disease: novel insights and future directions. EBioMedicine 90, 104511 (2023). 9. Iqbal, K. & Grundke-Iqbal, I. Alzheimer Disease, a Multifactorial Disorder Seeking Multi-therapies. Alzheimers Dement 6, 420 (2010). 10. Hampel, H. et al. The foundation and architecture of precision medicine in neurology and psychiatry. Trends Neurosci 46, 176-198 (2023). 11. Hebert, L. E., Weuve, J., Scherr, P. A. & Evans, D. A. Alzheimer disease in the United States (2010-2050) estimated using the 2010 census. Neurology 80, 1778-1783 (2013). 12. Jicha, G. A., Tucker, T. C., Arnold, S. M. & Nelson, P. T. Cancer research provides a model for advancing clinical trials in dementia in the era of disease-modifying Alzheimer’s-type dementia therapies. Alzheimers Res Ther 16, 184 (2024). 13. Higashi, S. et al. Concurrence of TDP-43, tau and a-synuclein pathology in brains of Alzheimer’s disease and dementia with Lewy bodies. Brain Res 1184, 284-294 (2007). 14. Meneses, A. et al. TDP-43 Pathology in Alzheimer’s Disease. Molecular Neurodegeneration 2021 16:1 16, 1-15 (2021). 15. Karanth, S. et al. Prevalence and Clinical Phenotype of Quadruple Misfolded Proteins in Older Adults. JAMA Neurol 77, 1299-1307 (2020). 16. Buratti, E. & Baralle, F. E. Characterization and functional implications of the RNA binding properties of nuclear factor TDP-43, a novel splicing regulator of CFTR exon 9. J Biol Chem 276, 36337-36343 (2001). 17. Neumann, M. et al. Ubiquitinated TDP-43 in Frontotemporal Lobar Degeneration and Amyotrophic Lateral Sclerosis. Science (American Association for the Advancement of Science) 314, 130-133 (2006). 18. Ma, X. R. et al. TDP-43 represses cryptic exon inclusion in the FTD-ALS gene UNC13A. Nature 603, 124-130 (2022). 19. Brown, A. L. et al. TDP-43 loss and ALS-risk SNPs drive mis-splicing and depletion of UNC13A. Nature 603, 131-137 (2022). 20. Klim, J. R. et al. ALS-implicated protein TDP-43 sustains levels of STMN2, a mediator of motor neuron growth and repair. Nat Neurosci 22, 167-179 (2019). 21. Melamed, Z. et al. Premature polyadenylation-mediated loss of stathmin-2 is a hallmark of TDP-43-dependent neurodegeneration. Nat Neurosci 22, 180-190 (2019). 22. Prudencio, M. et al. Truncated stathmin-2 is a marker of TDP-43 pathology in frontotemporal dementia. J Clin Invest 130, 6080-6092 (2020). 23. Dunker, W. et al. TDP-43 prevents endogenous RNAs from triggering a lethal RIG-I-dependent interferon response. Cell Rep 35, (2021). 24. Saldi, T. K. et al. TDP-1, the Caenorhabditis elegans ortholog of TDP-43, limits the accumulation of double-stranded RNA. EMBO J 33, 2947-2966 (2014). 25. LaRocca, T. J., Mariani, A., Watkins, L. R. & Link, C. D. TDP-43 knockdown causes innate immune activation via protein kinase R in astrocytes. Neurobiol Dis 132, (2019). 26. Rodriguez, S. et al. Machine learning identifies candidates for drug repurposing in Alzheimer’s disease. Nature Communications 2021 12:1 12, 1-13 (2021). 27. Sun, M. et al. Cryptic exon incorporation occurs in Alzheimer’s brain lacking TDP-43 inclusion but exhibiting nuclear clearance of TDP-43. Acta Neuropathol 133, 923-931 (2017). 28. Estades Ayuso, V. et al. TDP-43-regulated cryptic RNAs accumulate in Alzheimer’s disease brains. Mol Neurodegener 18, (2023). 29. Chang, K. et al. Loss of TDP-43 splicing repression occurs early in the aging population and is associated with Alzheimer’s disease neuropathologic changes and cognitive decline. Acta Neuropathol 147, (2023). 30. Agra Almeida Quadros, A. R. et al. Cryptic splicing of stathmin-2 and UNC13A mRNAs is a pathological hallmark of TDP-43-associated Alzheimer’s disease. Acta Neuropathol 147, (2024). 31. Chung, M. et al. Cryptic exon inclusion is a molecular signature of LATE-NC in aging brains. Acta Neuropathol 147, (2024). 32. Calliari, A. et al. HDGFL2 cryptic proteins report presence of TDP-43 pathology in neurodegenerative diseases. Mol Neurodegener 19, (2024). 33. Hoy, S. M. Deucravacitinib: First Approval. Drugs 82, 1671-1679 (2022). 34. Lin, J. R. et al. Highly multiplexed immunofluorescence imaging of human tissues and tumors using t-CyCIF and conventional optical microscopes. Elife 7, (2018). 35. Pereira, J. D. et al. Human sensorimotor organoids derived from healthy and amyotrophic lateral sclerosis stem cells form neuromuscular junctions. Nat Commun 12, (2021). 36. Marques, C. et al. Neuronal STING activation in amyotrophic lateral sclerosis and frontotemporal dementia. Acta Neuropathol 147, (2024). 37. De Jager, P. L. et al. A multi-omic atlas of the human frontal cortex for aging and Alzheimer’s disease research. Scientific Data 2018 5:1 5, 1-13 (2018). 38. Wang, M. et al. The Mount Sinai cohort of large-scale genomic, transcriptomic and proteomic data in Alzheimer’s disease. Scientific Data 2018 5:1 5, 1-16 (2018). 39. Echavarri, C. et al. Atrophy in the parahippocampal gyrus as an early biomarker of Alzheimer’s disease. Brain Struct Funct 215, 265 (2011). 40. Braak, H. & Braak, E. Neurofibrillary changes confined to the entorhinal region and an abundance of cortical amyloid in cases of presenile and senile dementia. Acta Neuropathol 80, 479-486 (1990). 41. Udeochu, J. C. et al. Tau activation of microglial cGAS-IFN reduces MEF2C-mediated cognitive resilience. NatNeurosci 26, (2023). 42. Braak, H. & Braak, E. Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 82, 239-259 (1991). 43. Liu, E. Y. et al. Loss of Nuclear TDP-43 Is Associated with Decondensation of LINE Retrotransposons. Cell Rep 27, 1409-142 Le6 (2019). 44. Ling, S. C., Polymenidou, M. & Cleveland, D. W. Converging mechanisms in als and FTD: Disrupted RNA and protein homeostasis. Neuron 79, 416-438 (2013). 45. Lee, P. L. et al. Posterior Cingulate Cortex Network Predicts Alzheimer’s Disease Progression. Front Aging Neurosci 12, 466 (2020). 46. Song, Y. et al. A dynamic view of the proteomic landscape during differentiation of ReNcell VM cells, an immortalized human neural progenitor line. Sci Data 6, (2019). 47. Fridman, J. S. et al. Selective Inhibition of JAKI and JAK2 Is Efficacious in Rodent Models of Arthritis: Preclinical Characterization of INCB028050. The Journal of Immunology 184, 5298-5307 (2010). 48. Quintas-Cardama, A. et al. Preclinical characterization of the selective JAK1 / 2 inhibitor INCB018424: therapeutic implications for the treatment of myeloproliferative neoplasms. Blood 115, 3109-3117 (2010). 49. Choi, S. H. et al. A three-dimensional human neural cell culture model of Alzheimer’s disease. Nature 515, 274-278 (2014). 50. Sanson, K. R. et al. Optimized libraries for CRISPR-Cas9 genetic screens with multiple modalities. Nature Communications 2018 9:1 9, 1-15 (2018). 51. Burke, J. R. et al. Autoimmune pathways in mice and humans are blocked by pharmacological stabilization of the TYK2 pseudokinase domain. Sci Transl Med 11, (2019). 52. Chimalakonda, A. et al. Selectivity Profile of the Tyrosine Kinase 2 Inhibitor Deucravacitinib Compared with Janus Kinase 1 / 2 / 3 Inhibitors. Dermatol Ther (Heidelb) 11, 1763-1776 (2021). 53. Sun, B. B. et al. Plasma proteomic associations with genetics and health in the UK Biobank. Nature 622, 329-338 (2023). 54. Ferkingstad, E. et al. Large-scale integration of the plasma proteome with genetics and disease. Nat Genet 53, 1712-1721 (2021). 55. Dendrou, C. A. et al. Resolving TYK2 locus genotype-to-phenotype differences in autoimmunity. Sci Transl Med 8, (2016). 56. Kreins, A. Y. et al. Human TYK2 deficiency: Mycobacterial and viral infections without hyper-IgE syndrome. J Exp Med 212, 1641-1662 (2015). 57. Boisson-Dupuis, S. et al. Tuberculosis and impaired IL-23-dependent IFN-y immunity in humans homozygous for a common TYK2 missense variant. Sci Immunol 3, (2018). 58. Verite, J. et al. Peripheral Blood Mononuclear Cells of Alzheimer’s Disease Patients Control CCL4 and CXCL10 Levels in a Human Blood Brain Barrier Model. Curr Alzheimer Res 14, (2017). 59. Koper, O. M., Kaminska, J., Sawicki, K. & Kemona, H. CXCL9, CXCL10, CXCL11, and their receptor (CXCR3) in neuroinflammation and neurodegeneration. Adv Clin Exp Med 27, 849-856 (2018). 60. Correa, J. D., Starling, D., Teixeira, A. L., Caramelli, P. & Silva, T. A. Chemokines in CSF of Alzheimer’s disease patients. Arq Neuropsiquiatr 69, 455-459 (2011). 61. Barton, S. K. et al. Mutant TDP-43 Expression Triggers TDP-43 Pathology and Cell Autonomous Effects on Primary Astrocytes: Implications for Non-cell Autonomous Pathology in ALS. Neurochem Res 45, 1451-1459 (2020). 62. van Eersel, J. et al. Cytoplasmic accumulation and aggregation of TDP-43 upon proteasome inhibition in cultured neurons. PLoS One 6, (2011). 63. Li, X. et al. Systematic Analysis and Biomarker Study for Alzheimer’s Disease. Sci Rep 8, (2018). 64. Wan, J. et al. Tyk2 / STAT3 signaling mediates beta-amyloid-induced neuronal cell death: implications in Alzheimer’s disease. JNeurosci 30, 6873-6881 (2010). 65. Nevado-Holgado, A. J. et al. Genetic and Real-World Clinical Data, Combined with Empirical Validation, Nominate Jak-Stat Signaling as a Target for Alzheimer’s Disease Therapeutic Development. Cells 8, (2019). 66. Dorrity, T. J. et al. Long 3’UTRs predispose neurons to inflammation by promoting immunostimulatory double-stranded RNA formation. Sci Immunol 8, (2023). 67. Sanders, O. D. Virus-Like Cytosolic and Cell-Free Oxidatively Damaged Nucleic Acids Likely Drive Inflammation, Synapse Degeneration, and Neuron Death in Alzheimer’s Disease. J Alzheimers Dis Rep 7, 1-19 (2023). 68. Melton, L. M. et al. Chronic glial activation, neurodegeneration, and APP immunoreactive deposits following acute administration of double-stranded RNA. Glia 44, 112 (2003). 69. Milstead, R. A., Link, C. D., Xu, Z. & Hoeffer, C. A. TDP-43 knockdown in mouse model of ALS leads to dsRNA deposition, gliosis, and neurodegeneration in the spinal cord. Cereb Cortex (2022) doi:10.1093 / CERCOR / BHAC461. 70. Frost, B. & Dubnau, J. The Role of Retrotransposons and Endogenous Retroviruses in Age-Dependent Neurodegenerative Disorders. Annu Rev Neurosci 47, (2024). 71. Gorbunova, V. et al. The role of retrotransposable elements in ageing and age-associated diseases. Nature 596, 43-53 (2021). 72. Grochowska, J., Czerwinska, J., Borowski, L. S. & Szczesny, R. J. Mitochondrial RNA, a new trigger of the innate immune system. Wiley Interdiscip Rev RNA 13, (2022). 73. Biagio, P., Isabella, D. F., Federica, C., Elena, S. & Ivan, G. Alzheimer’s disease and herpes viruses: Current events and perspectives. Rev Med Virol 34, e2550 (2024). 74. Farhangian, M., Azarafrouz, F., Valian, N. & Dargahi, L. The role of interferon beta in neurological diseases and its potential therapeutic relevance. Eur J Pharmacol 176882 (2024) doi: 10.1016 / J.EJPHAR.2024.176882. 75. Chhatbar, C. et al. Type I Interferon Receptor Signaling of Neurons and Astrocytes Regulates Microglia Activation during Viral Encephalitis. Cell Rep 25, 118-129.e4 (2018). 76. Ferecsko, A. S. et al. STING-Triggered CNS Inflammation in Human Neurodegenerative Diseases. Biomedicines 11, (2023). 77. Hoisnard, L. et al. Adverse events associated with JAK inhibitors in 126,815 reports from the WHO pharmacovigilance database. Scientific Reports 2022 12:1 12, 1-10 (2022). 78. Pellenz, F. M. et al. Association of TYK2 polymorphisms with autoimmune diseases: A comprehensive and updated systematic review with meta-analysis. Genet Mol Biol 44, (2021). 79. Dendrou, C. A. et al. Resolving TYK2 locus genotype-to-phenotype differences in autoimmunity. Sci Transl Med 8, (2016). 80. Papp, K. et al. Phase 2 Trial of Selective Tyrosine Kinase 2 Inhibition in Psoriasis. N Engl J Med 379, 1313-1321 (2018). 81. Wrobleski, S. T. et al. Highly Selective Inhibition of Tyrosine Kinase 2 (TYK2) for the Treatment of Autoimmune Diseases: Discovery of the Allosteric Inhibitor BMS-986165. J Med Chem 62, 8973-8995 (2019). 82. Seddighi, S. et al. Mis-spliced transcripts generate de novo proteins in TDP-43 -related ALS / FTD. bioRxiv 2023.01.23.525149 (2023) doi: 10.1101 / 2023.01.23.525149. 83. Irwin, K. E. et al. A fluid biomarker reveals loss of TDP-43 splicing repression in pre-symptomatic ALS. bioRxiv 2023.01.23.525202 (2023) doi:10.1101 / 2023.01.23.525202. 84. Nicks, R. et al. Repetitive head impacts and chronic traumatic encephalopathy are associated with TDP-43 inclusions and hippocampal sclerosis. Acta Neuropathol 145, (2023). 85. Ochoa, E. et al. Pathogenic tau-induced transposable element-derived dsRNA drives neuroinflammation. Sci Adv 9, (2023). 86. Chang, Y. H. & Dubnau, J. Endogenous retroviruses and TDP-43 proteinopathy form a sustaining feedback driving intercellular spread of Drosophila neurodegeneration. Nat Commun 14, (2023). 87. Li, Q. et al. RNA editing underlies genetic risk of common inflammatory diseases. Nature 2022 608:7923 608, 569-577 (2022). 88. Levine, K. S. et al. Virus exposure and neurodegenerative disease risk across national biobanks. Neuron 111, (2023). 89. Jamieson, G. A., Maitland, N. J., Wilcock, G. K., Craske, J. & Itzhaki, R. F. Latent herpes simplex virus type 1 in normal and Alzheimer’s disease brains. J Med Virol 33, 224227(1991). Nariya, M. K., Mills, C. E., Sorger, P. K. & Sokolov, A. Paired evaluation of machine-learning models characterizes effects of confounders and outliers. Patterns 4, (2023). 91. Bihlmeyer, N. A. et al. Novel methods for integration and visualization of genomics and genetics data in Alzheimer’s disease. Alzheimers Dement 15, 788 (2019). 92. Patel, A., Garcia Diaz, A., Moore, J. C., Sirabella, D. & Comeo, B. Establishment and characterization of two iPSC lines derived from healthy controls. Stem Cell Res (2020). 93. Moakley, D. et al. Pharmacological Profding of Purified Human Stem Cell-Derived and Primary Mouse Motor Neurons. Sci Rep 9, (2019). 94. Song, Y. et al. A dynamic view of the proteomic landscape during differentiation of ReNcell VM cells, an immortalized human neural progenitor line. Scientific Data 2019 6:1 6, 1-17(2019). 95. Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9, 671-675 (2012). 96. Patro, R., Duggal, G., Love, M. I., Irizarry, R. A. & Kingsford, C. Salmon provides fast and bias-aware quantification of transcript expression. Nature Methods 201714:4 14, 417-419(2017). 97. Soneson, C., Love, M. I. & Robinson, M. D. Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences. FlOOOResearch 2016 4:1521 4, 1521 (2016). 98. Airola, A., Pahikkala, T., Waegeman, W., De Baets, B. & Salakoski, T. An experimental comparison of cross-validation techniques for estimating the area under the ROC curve. Comput Stat Data Anal 55, 1828-1844 (2011). 99. Doench, J. G. et al. Optimized sgRNA design to maximize activity and minimize off-target effects of CRISPR-Cas9. Nature Biotechnology 2015 34:2 34, 184-191 (2016). 100. Piccioni, F., Younger, S. T. & Root, D. E. Pooled Lenti viral-Delivery Genetic Screens. Curr ProtocMol Biol 121, 32.1.1-32.1.21 (2018). 101. Rappsilber, J., Ishihama, Y. & Mann, M. Stop and Go Extraction Tips for Matrix-Assisted Laser Desorption / Ionization, Nanoelectrospray, and LC / MS Sample Pretreatment in Proteomics. Anal Chem 75, 663-670 (2002). 102. McAlister, G. C. et al. MultiNotch MS3 enables accurate, sensitive, and multiplexed detection of differential expression across cancer cell line proteomes. Anal Chem 86, 71507158 (2014). 103. Paulo, J. A. et al. Quantitative mass spectrometry-based multiplexing compares the abundance of 5000 S. cerevisiae proteins across 10 carbon sources. J Proteomics 148, 85-93 (2016). 104. Paulo, J. A., O’Connell, J. D. & Gygi, S. P. A Triple Knockout (TKO) Proteomics Standard for Diagnosing Ion Interference in Isobaric Labeling Experiments. J Am Soc Mass Spectrom 27, 1620-1625 (2016). 105. Loh, P. R. et al. Efficient Bayesian mixed-model analysis increases association power in large cohorts. Nat Genet 47, 284-290 (2015). 106. Perez-Riverol, Y. et al. The PRIDE database resources in 2022: a hub for mass spectrometry-based proteomics evidences. Nucleic Acids Res 50, D543-D552 (2022). OTHER EMBODIMENTS It is to be understood that while the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims. Other aspects, advantages, and modifications are within the scope of the following claims.

Claims

1. A method of administering a smell test for assessment of risk of dementia in a subject comprising:(a) generating, by a computing device, a user interface to be displayed on the computing device;(b) instructing, by the computing device via the user interface, the subject to utilize an odor card configured to deliver an odor from among a plurality of odors;(c) instructing, by the computing device via the user interface, the subject to identify the delivered odor from a forced choice list of options provided by the computing device via the user interface;(d) asking, by the computing device via the user interface, the subject to evaluate their confidence in their odor identification decision;(e) receiving, by the computing device via the user interface, a subject input for a smell test;(f) calculating, by the computing device, a confidence metric for the smell test; and(g) identifying the subject as high risk for dementia based on the confidence metric.

2. A method comprising:(a) instructing a subject to utilize an odor card configured to deliver an odor from among a plurality of odors;(b) instructing the subject to identify the delivered odor from a forced choice list of options;(c) asking the subject to evaluate their confidence in their odor identification decision; and(d) calculating, optionally by a computing device, a confidence metric based on the subject’s confidence in their odor identification decision.

3. The method of claim 1 or 2, wherein the odor card is configured to deliver an odor from among a plurality of nine odors.

4. The method of claim 3, wherein the plurality of nine odors comprises at least at least two, three, four, five or more of, or consists of all of, menthol, clove, leather, strawberry, lilac, pineapple, smoke, soap, and grape.

5. The method of any one of claims 1-4, wherein the smell test further includes an odor episodic memory test.

6. The method of claim 1, further comprising administering an odor episodic memory test, wherein the test comprises:(li) instructing, by the computing device via the user interface, the subject to utilize a second odor card configured to deliver an odor from among a plurality of odors;(i) asking, by the computing device via the user interface, the subject if the odor delivered in the second odor card was delivered in the earlier used odor card (i.e., a first odor card);(j) instructing, by the computing device via the user interface, the subject to identify the delivered odor from a forced choice list of options provided by the computing device via the user interface;(k) asking, by the computing device via the user interface, the subject to evaluate their confidence in their odor identification decision;(1) receiving, by the computing device via the user interface, a subject input for a smell test;(m) calculating, by the computing device, a confidence metric for the smell test; and(n) identifying the subject as high risk for dementia based on the confidence metric.

7. The method of claim 6, wherein the odor card is configured to deliver an odor from among a plurality of eighteen odors.

8. The method of claim 7, wherein the plurality of eighteen odors comprises at least two, three, four, five or more of, or consists of all of, menthol, clove, leather,strawberry, lilac, pineapple, smoke, soap, grape, coffee, peach, chocolate, orange, dirt, banana, lemon, bubble gum, and rose.

9. The method of any one of claims 1-8, wherein the asking, by the computing device via the user interface, prompts the subject to evaluate their confidence in their odor identification decision by selecting one answer from the following “I Guessed,” “I Narrowed Down to Three,” “I Narrowed Down to Two,” or “I Am Certain.”10. The method of any one of claims 1-9, wherein the user interface is a website or a web-based application.

11. The method of any one of claims 1-10, wherein the computing device is a computer, a smart phone, or a mobile device.

12. The method of any one of claims 1-11, wherein the method is selfadministered by the subject.

13. The method of any one of claims 1-12, wherein the confidence metric for the smell test is calculated by the computing device as the total number of odors identified correctly by the subject where the subject did not evaluate their confidence in their odor identification decision by selecting “I Guessed” when prompted by the computing device via the user interface.

14. The method of any one of claims 1-13, further comprising comparing, by the computing device, the calculated confidence metric for the smell test relative to a predetermined threshold, wherein the predetermined threshold is determined by a machine learning algorithm based on a plurality of previously entered smell test performance data from healthy subjects.

15. The method of any one of claims 1-14, wherein the subject is identified as high risk for dementia when the calculated confidence metric for the smell test is lower than the predetermined threshold.

16. A method comprising:(a) collecting a biological sample from a subject, preferably a biological sample comprising whole blood, serum, or plasma;(b) quantifying a level of at least one biomarker for dementia in the biological sample, wherein the at least one biomarker for dementia is selected from the group consisting of C-X-C motif chemokine ligand 10 (CXCL10), C-C motif chemokine ligand 2 (CCL2), and interleukin 6 (IL-6); and(c) comparing the level of the least one biomarker for dementia to a predetermined threshold, wherein the predetermined threshold is determined from levels of the biomarker assessed from biological samples collected from a plurality of healthy subjects.

17. The method of claim 16, further comprising identifying the subject as at high risk for dementia when the level of at least one biomarker for dementia is higher than the predetermined threshold.

18. The method of claim 16, further comprising identifying the subject as at high risk for dementia when the level of at least two biomarkers for dementia are higher than the predetermined thresholds.

19. The method of claim 16, further comprising identifying the subject as at high risk for dementia when the level of CXCL10, CCL2, IL-6, or any combination thereof is higher than the predetermined threshold.

20. A method of assessing the risk of dementia in a subject comprising administering the smell test according to the method of any one of claims 1-15 in combination with the method of any one of claims 16-19.

21. A method of treating a subject for dementia, wherein the subject is at high risk for dementia as determined by the smell test according to the method of any one of claims 1-15, the method of any one of claims 16-19, or a combination thereof.

22. The method of claim 21, wherein the treatment comprises administering to thesubject at least one treatment for Alzheimer’s disease, optionally wherein the treatment is selected from the group consisting of benzgalantamine, donepezil, galantamine, rivastigmine, memantine, lecanemab, donanemab, suvorexant, and brexpiprazole.

23. The method of claim 21, wherein the treatment comprises administering to the subject at least one therapeutic agent targeting TYK2.

24. The method of claim 23, wherein the at least one therapeutic agent targeting TYK2 is a TYK2 inhibitor, optionally wherein the TYK2 inhibitor is deucravacitinib.

25. A method of administering a smell test for assessment of neuroanatomical volume in at least one brain region in a subject comprising:(a) administering the smell test according to the methods of any one of claims 1-15; and(b) identifying a decline in neuroanatomical volume in at least one brain region if the calculated confidence metric for the smell test is lower than the predetermined threshold.

26. A method of administering a smell test for assessment of neuroanatomical volume in at least one brain region in a subject comprising:(a) generating, by a computing device, a user interface to be displayed on the computing device;(b) instructing, by the computing device via the user interface, the subject to utilize an odor card configured to deliver an odor from among a plurality of odors;(c) instructing, by the computing device via the user interface, the subject to identify the delivered odor from a forced choice list of options provided by the computing device via the user interface;(d) asking, by the computing device via the user interface, the subject to evaluate their confidence in their odor identification decision;(e) receiving, by the computing device via the user interface, a subject input for a smell test;(f) calculating, by the computing device, a score for the smell test; and(g) identifying a decline in neuroanatomical volume in at least one brain region in the subject based on the score for the smell test.

27. The method of claim 26, wherein the score for the smell test is calculated as a total number of odors correctly identified by the subject.

28. The method of claim 26 or 27, wherein the odor card is configured to deliver an odor from:(i) among a plurality of nine odors, optionally wherein the plurality of nine odors comprises at least at least two, three, four, five or more of, or consists of all of, menthol, clove, leather, strawberry, lilac, pineapple, smoke, soap, and grape; and / or,(ii) from among a plurality of eighteen odors, optionally wherein the plurality of eighteen odors comprises at least two, three, four, five or more of, or consists of all of, menthol, clove, leather, strawberry, lilac, pineapple, smoke, soap, grape, coffee, peach, chocolate, orange, dirt, banana, lemon, bubble gum, and rose.

29. The method of any one of claims 26-28, wherein the user interface is a website or a web-based application.

30. The method of any one of claims 26-29, wherein the computing device is a computer, a smart phone, or a mobile device.

31. The method of any one of claims 26-30, wherein the method is selfadministered by the subject.

32. The method of any one of claims 26-31, further comprising comparing, by the computing device, the calculated score for the smell test relative to a predetermined threshold, wherein the predetermined threshold is determined from smell test performance data collected from a plurality of healthy subjects.

33. The method of any one of claims 25-32, wherein the at least one brain region in the subject comprises hippocampus, amygdala, and / or both.

34. The method of claim 33, wherein the at least one brain region in the subject comprises the left hippocampus, the right hippocampus, the left amygdala, the right amygdala, or any combination thereof.

35. The method of any one of claims 25-34, wherein the identification of decline in neuroanatomical volume in at least one brain region in the subject correlates to cognitive decline in the subject.

36. The method of any one of claims 1-35, wherein the subject has, is suspected of having, or is at risk of having a neurodegenerative disease associated with TDP-43 pathology.

37. The method of any one of claims 1-36, wherein the subject has, is suspected of having, or is at risk of having Alzheimer’s disease.

38. The method of claim 37, wherein the subject at risk of having Alzheimer’s disease has a loss-of-function SNP in TYK2 (rs34536443).

39. A method for selecting a subject for treatment with a TYK2 inhibitor, the method comprising:(a) administering the smell test according to the methods of any one of claims 1-15 and 25-39; and(b) selecting the subject for treatment with a TYK2 inhibitor if the calculated confidence metric for the smell test is lower than the predetermined threshold.

40. The method of claim 39, optionally further comprising quantifying a level of at least one biomarker for dementia in a biological sample collected from the subject according to the method of any one of claims 16-19, and selecting the subject for treatment with a TYK2 inhibitor if the level of at least one biomarker for dementia is higher than the predetermined threshold.

41. The method of claim 39 or 40, further comprising (c) administering the TYK2 inhibitor to the subject if the subject is selected for treatment.

42. The method of any one of claims 39-41, wherein the TYK2 inhibitor is deucravacitinib.

43. The method of any one of claims 39-42, wherein the subject has, is suspected of having, or is at risk of having a neurodegenerative disease associated with TDP-43 pathology.

44. The method of claim 43, wherein the neurodegenerative disease associated with TDP-43 pathology comprises a subset of Alzheimer’s disease.

Images