Computerized method of endotyping to assess alzheimer's disease progression and personalized treatment thereof

Abstract

Provided herein in certain embodiments is an Al-assisted method based to assess progression of mild cognitive impairment or a risk of developing Alzheimer's disease (AD) based on proteomic and metabolomic data of one or more samples obtained from a human subject via a portal. The novel portal described herein is based on a trained computational model for assignment of a dataset into three endotypes that is comprehensive enough to provide accurate assessments of AD risk and respective treatments for each, yet simple enough to avoid excessive energy usage demand on clinics and hospitals.

Description

COMPUTERIZED METHOD OF ENDOTYPING TO ASSESS ALZHEIMER’S DISEASE PROGRESSION AND PERSONALIZED TREATMENT THEREOF TECHNICAL FIELD

[0001] The present disclosure relates generally to the field of computer-implemented assessment of subject datasets to assess Alzheimer’s disease progression.BACKGROUND

[0002] Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by cognitive decline, memory loss and behavioral changes. Early detection and intervention are critical to mitigating the impact of AD and improving patient outcomes. Traditional diagnostic methods often fall short of identifying individuals at risk of transitioning from mild cognitive impairment (MCI) to AD, especially since cognitive impairment can be difficult to measure. Furthermore, it can be challenging to rule out other disorders that cause similar symptoms, such as thyroid disorders and vitamin B-12 deficiency. Currently, physicians use brain imaging or mental status tests to determine the degree of cognitive impairment. While researchers are examining more accurate ways to diagnose Alzheimer’s disease (including assessing biomarkers such as tau), progress in the field has been slow.

[0003] A further challenge is that Alzheimer’s results from multiple disease mechanisms that can vary from individual to individual. The biomarker information generated from high throughput spectroscopic analysis of patient samples is vast, making it impractical for humans to interpret. While it may be possible to analyze genomic data, metabolomic and proteomic data is far more complex. Handling, analyzing, and interpreting this complex biological data requires advanced, non-conventional computational tools, which increases the need for data storage and computational power.

[0004] Accordingly, while understanding the biological pathways in which biomarkers operate could improve personalized treatment, elucidating such pathways from complex datasets that cannot be practically performed by a human still remains a challenge. A study examining molecular signatures of Alzheimer’s disease from metabolomic or proteomic data sets analyzed separately revealed various, complicated deregulated pathways which included neurotransmittersynapses, oxidative stress, inflammation, vitamins, complement and coagulation pathways (Kodam et al., 2023, Nature, 13:3695). Other investigators have identified cell death, cellular senescence, energy metabolism, genomic integrity, glia, immune system, metal ion homeostasis, oxidative stress, proteostasis, and synaptic function using (Shokhirev and Johnson, 2022, Ageing Research Reviews, 81:101721). While this information proves useful for understanding the various pathways that contribute to the disease, there is currently no approach to assign disease risk that could be easily applied in practical settings, such as a clinic or hospital.

[0005] Furthermore, while these methods rely on machine learning in a research setting to elucidate the pathways, in a practical setting, the computing power to assess such multiple, complicated pathways for disease progression in a subject may require significant data center usage. Data centers consume significant energy to run and cool servers with consequent emission of greenhouse gases. An exponentially increasing demand for computational capacity to power Al training will have negative implications for the environment.

[0006] Hospitals and medical clinics are large consumers of energy. In a clinical setting, energyconsuming technologies, such as electronic imaging equipment and digital record keeping add to the existing energy load. Increased data usage can result in delays in response times when accessing computer data as well as limited bandwidth to access cloud-based applications. As novel medical devices are introduced into clinics, they need to be connected to not only the hospital but also the cloud.

[0007] While elucidating multiple, complex pathways to assess disease progression is the current approach used by machine learning methodologies, other approaches attribute Alzheimer’s disease progression to single categories of mechanisms. For example, inflammation has been identified as a central mechanism in Alzheimer’s disease (Kinney et al., 2018, Alzheimer’s and Dementia: Translational Research & Clinical Interventions, 4:575-590). By contrast, metabolic dysregulation has been considered responsible for disease progression (Yan et al., 2020, Frontiers in Neuroscience, volume 14). In further studies, sphingolipids have been identified as playing a key role in AD (deWit et al., 2021, Frontiers in Immunology, 11:620348).

[0008] As a result, current methods to assess disease progression still rely on cognitive tests or medical imaging data. More recently it has been demonstrated that machine learning cancontribute to the analysis of neuroimaging data in dementia care. BrainSee™ is an FDA approved artificial intelligence-based procedure that delivers a quantitative score from 0 to 100, allowing for patients to be categorized by risk when considering potential treatment options. However, neuroimaging is not easily accessible for all patients due to its cost or accessibility.

[0009] Furthermore, Alzheimer’s disease may occur simultaneously with other chronic diseases or medical conditions. Such comorbidities may be undetected by conventional means and are often pre-symptomatic. Patient outcomes could be improved by identifying other health conditions that could otherwise complicate management of a subject’s overall health.

[0010] Accordingly, an improved computerized method is needed that can process vast amounts of biomarker data generated from high throughput screening and assess the progression of Alzheimer’s disease, without overconsumption of data center usage.SUMMARY

[0011] The present disclosure relates generally to the field of computer-implemented assessments of subject datasets to assess Alzheimer’s disease progression.

[0012] According to the present disclosure, in some embodiments, there is provided a computer-implemented method that improves digital monitoring of the progression or regression of mild cognitive impairment (MCI) to Alzheimer’s disease (AD) or the progression or regression of Alzheimer’ s disease from large biomarker datasets (e.g., greater than 50 biomarkers). The method is based on a machine learning computation model, which is trained on a novel dataset of proteomic and metabolomic data obtained from subjects with neurological conditions, including dementia, mild cognitive impairment (MCI) and AD. Advantageously, the machine learning model developed by the inventors is used to identify three distinct AD endotypes from the vast datasets of metabolomic and proteomic biomarkers that otherwise could not be practically assessed by a human. Assigning a sample to one or more of the three endotypes identified by the machine learning model improves AD risk assessment. Since a subject may fall into only one of the three endotypes and not others, the method can also avoid false negatives that might otherwise occur with methods that rely on identifying risk based on a single disease mechanism (e.g., examining only inflammation, sphingolipid and metabolic dysregulation rather than all three).

[0013] The ability to assess a subject’s risk of developing AD before they are symptomatic or have mild cognitive impairment (MCI) by the inventive, computer-implemented endo-typing method also enables more targeted, personalized treatments for slowing or potentially reversing AD progression. For example, by identifying a sub-set of markers within each endotype, treatment can be aimed at adjusting the biomarkers levels using drugs, supplements, etc. and monitoring those biomarkers falling within the assigned endotype over time, thereby improving patient care.

[0014] Furthermore, the approach of stratifying subjects into at least one of three distinct endotypes is simpler to implement than previous approaches that rely on the identification of multiple, complex disease etiologies responsible for Alzheimer’s disease progression.

[0015] The method of the disclosure also provides a simplified training model, thereby improving the function of the computer, without compromising on predictive power. Improved computing efficiency by avoiding the assessment of multiple, complicated pathways as per prior methods, reduces data center usage or energy efficiency, which has become an ongoing problem in clinical settings as new digital technologies are adopted.

[0016] In some embodiments, the disclosure relies on machine learning of a data set by multimodal batch processing. This single model is designed to simultaneously identify and classify the endotypes. By processing a batch of patient samples at once, the model can leverage the shared computational resources and avoid the redundant calculations that would be necessary for running three separate models. Further, unlike standard mini-batching that processes homogeneous data of a single modality, the disclosed multimodal batch processing operates on synchronized data from multiple modalities within each batch, enabling concurrent cross-modality learning and improved computational efficiency.

[0017] In summary, the inventors have developed a portal derived from improved machine learning that is comprehensive enough to provide accurate assessments of AD risk and respective treatments for each, yet simple enough to avoid excessive energy usage demand on clinics and hospitals.

[0018] In addition, a unique subject portal is disclosed herein that facilitates the identification of treatments assigned to each of the three endotypes and in some embodiments provides an improved interface and sharing of data between two parties, such as a health care provider and a subject. Such portal in some embodiments is capable of assigning comorbidities, allows for patient and caregiver access and / or allows for ease of downloading of subject data in a convenient digital format.SUMMARY

[0019] Figure 1A.1 provides a listing of biomarkers that can contribute to progression of mild cognitive impairment to AD.

[0020] Figure 1A.2 provides a listing of biomarkers that can contribute to progression of mild cognitive impairment to AD.

[0021] Figure 1B shows the three AD endotypes described herein that were identified by the computer model.

[0022] Figure 1C shows the number of biomarkers that were out of range for each subject assigned one of the three endotypes.

[0023] Figure 1D depicts how a sample is stratified in patients using the computer model by identifying clear patterns of AD biomarkers in data from a first blood draw.

[0024] Figure 2A shows amino acid, complement system proteins, sphingolipid and lysphospholipid biomarker data from a full set of patient samples in a first study processed by the computer model. The darkened boxes are indicative of biomarkers that were out of range as identified by the computer model.

[0025] Figure 2B shows amino acid, complement system proteins, sphingolipid and lysphospholipid biomarker data from patient samples. The darkened boxes are indicative of biomarkers that were out of range as identified by the computer model. The data is a sub-set of samples of Figure 2 A.

[0026] Figure 2C shows amino acid, other amino acids and derivatives, and inflammation and oxidative stress biomarkers from patient samples of Figure 2A.

[0027] Figure 3 shows amino acid, complement system proteins, sphingolipid and lysphospholipid biomarker data from the patient samples processed by the trained computational model. The darkened boxes are indicative of biomarkers that were out of range as identified by the computer model.

[0028] Figure 4A depicts an example of an electronic portal showing a treatment for an endotype identified by the computer model.

[0029] Figure 4B depicts an example of an electronic portal showing instructions for sharing a sign-up link or to create a profile on behalf of a patient.

[0030] Figure 4C shows longitudinal tracking of biomarkers over time after a patient has been assigned an Alzheimer’s treatment schedule.

[0031] Figure 5 is a flowsheet showing a system for categorizing endotypes comprising training a computational model using multimodal batch processing on proteomic and metabolomic datasets to produce a trained computational model; measuring a plurality of proteomic and metabolomic biomarkers from a sample of a human subject via a high-throughput spectroscopic assay; and categorizing the plurality of biomarkers categorized into at least one of three endotype classes that cause disease progression.DETAILED DESCRIPTION

[0032] A detailed description of one or more embodiments of the invention is provided below along with accompanying figures that illustrate the principles of the invention. The invention is described in connection with such embodiments, but the invention is not limited to any particular embodiment described herein. The scope of the invention is limited only by the claims and equivalents thereof. Numerous specific details are set forth in the following description in order to provide a thorough understanding of the invention. It being understood that various changes can be made in the function and arrangement of elements without departing from the scope as set forth in the claims. Accordingly, an embodiment is an example or implementation of the inventions andnot the sole implementation. Various appearances of “one embodiment,” “an embodiment” or “some embodiments” do not necessarily all refer to the same embodiments. Although various features of the invention may be described in the context of a single embodiment, the features may also be provided separately or in any suitable combination. Conversely, although the invention may be described herein in the context of separate embodiments for clarity, the invention can also be implemented in a single embodiment or any combination of embodiments. These details are provided for the purpose of providing non-limiting examples and the invention may be practiced according to the claims without some or all of these specific details. For the purpose of clarity, certain technical material that is known in the technical fields related to the invention has not been described in detail so that the invention is not unnecessarily obscured by such descriptions.Definitions

[0033] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by a person of ordinary skill in the art to which the present invention pertains. As used herein, and unless stated otherwise or required otherwise by context, each of the following terms shall have the definition set forth below.

[0034] Articles such as “a” and “an” when used in a claim, are understood to mean one or more of what is claimed or described. It is to be understood that where the specification states that a component feature, structure, or characteristic “may”, “might”, “can” or “could” be included, that particular component, feature, structure, or characteristic is not required to be included.

[0035] The terms “comprises”, “comprising”, “include”, “includes”, “including”, “contain”, “contains” and “containing” are meant to be non-limiting, i.e., other steps and other sections which do not affect the end of result can be added. The above terms encompass the terms “consisting of and “consisting essentially of. These phrases, and grammatical variants thereof, when used herein are not to be construed as excluding additional components, steps, features integers or groups thereof but rather that the additional features, integers, steps, components or groups thereof do not materially alter the basic and novel characteristics of the claimed composition, device or method. If the specification or claims refer to “an additional” element, that does not preclude there being more than one of the additional element.

[0036] The term “normal range” with reference to classification of biomarker refers to a range of values obtained or derived from a plurality of individuals that do not suffer from Alzheimer’s disease.

[0037] The term “out of a normal range” or “out of range” with reference to a biomarker refers to a classification of a biomarker by a computational model in a sample that is at a concentration that falls outside of a normal range. The concentration can be higher or lower than an Alzheimer’ s-negative sample or a normal concentration range obtained from literature.

[0038] The term “treatment” or grammatical variations thereof generally refers to a treatment made in response to a subject sample assigned to at least one of the three assigned endotypes. The treatment may include, but is not limited to, one or more of the alleviation or prevention of symptoms, slowing or stopping the progression or worsening of Alzheimer’s and the remission of Alzheimer’s. In some embodiments, treatment refers to therapeutic treatment (e.g., changing biomarker levels associated with an endotype), which may include administration of a drug, supplements and / or nutrients to modify the concentration of a biomarker, e.g., as measured in a subsequent sample, so that it is within a normal range.

[0039] The term “causing treatment” or “causing treating” refers to a first party, such as a diagnostic provider, causing a second party, such as a health provider, to treat a subject based on one or more of the endotype(s) assigned to the sample by the computational model herein.

[0040] The term “endotype” refers to an AD risk subtype assigned to a subject sample based on a dataset of a subject’s metabolomic and proteomic profile as described herein.

[0041] The term “sphingolipid dysregulation” with reference to endotype 1 refers to levels of a sphingolipid biomarker, such as sphingomyelin or a ceramide, being out of a normal range in a sample.

[0042] The term “amino acid and metabolism dysregulation” with reference to an endotype refers to levels of amino acid or metabolite biomarkers being out of a normal range in a sample.

[0043] The term “inflammation and oxidative stress” with reference to an endotype refers to levels of inflammation and oxidative stress biomarkers being out of a normal range in a sample.

[0044] By the term “computational model”, it is meant an Al model that has been trained on metabolomic and proteomic data from patients with and without Alzheimer’s disease.

[0045] By the term “multimodal batch processing,” it is meant a computer-implemented operation in which metabolomic and proteomic data are processed concurrently within coordinated batches. Each batch comprises data elements that are synchronized across the modalities for a given set of samples or subjects.

[0046] The term “metabolite” generally refers to any molecule involved in metabolism. Metabolites can be products, substrates or intermediates in metabolic processes. Metabolites may include, without limitation, amino acids, peptides, acylcamitines, monosaccharides, lipids and phospholipids, lysophospholipid, sphingolipids, glycerophospholipids, glucose, prostaglandins, hydroxy eicosatetraenoic acids, hydroxy octadecadienoic acids, steroids, bile acids, and glycolipids and phospholipids.

[0047] The term “Alzheimer’ s-related metabolite” or “metabolomic profile” generally refers to metabolites associated with Alzheimer’s disease comprising one, or two or more metabolites described herein or a combination thereof.

[0048] The term “Alzheimer’ s-related protein” or “proteomic profile” generally refers to a profile of proteins, protein fragments and / or peptides associated with Alzheimer’s disease comprising two or more, three or more, four or more, or five or more proteins described herein or a combination thereof. As would be appreciated by those of skill in the art, quantification of a protein can comprise quantifying a fragment or peptide thereof.

[0049] The term “Alzheimer’ s-related protein” or “proteomic profile” generally refers to proteins associated with Alzheimer’s disease comprising one, or two or more proteins described herein or a combination thereof.

[0050] The terms “preferred”, “preferably” and variants generally refer to embodiments of the disclosure that afford certain benefits, under certain circumstances. However, other embodiments may also be preferred, under the same or other circumstances. Furthermore, the recitation of one or more preferred embodiments is not intended to exclude other embodiments from the scope of the disclosure.

[0051] The term “preventing”, and “prevention” are used interchangeably and generally refer to any activity that leads to a reduction in risk of developing Alzheimer’s disease in the subject.

[0052] The term “Alzheimer’ s-related protein” or “proteomic profile” generally refers to a profile of proteins, protein fragments and / or peptides associated with Alzheimer’s disease comprising two or more, three or more, four or more, or five or more proteins described herein or a combination thereof. As would be appreciated by those of skill in the art, quantification of a protein can comprise quantifying a fragment or peptide thereof.

[0053] A “marker” as used herein may refer to, but is not limited to, a biological marker (biomarker) which is a measurable indicator of some biological state or condition where the biomarker is often measured using blood, urine, or soft tissues of an individual to establish a level of the biomarker with respect to normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention for example. A biomarker may be genomic (genetic) marker, e.g., a specific gene or DNA sequence within a known location of a chromosome associated with having a disease or condition or likelihood of acquiring the disease or condition, a proteomic marker, also known as a protein biomarker relating to a protein or set of proteins associated with having the disease or condition or likelihood of acquiring the disease or condition, a metabolomic markers, relating to one or more small molecules commonly referred to as metabolites within cells, biofluids, tissues etc. or the chemical fingerprint left behind by a specific cellular process associated with having the disease or condition or likelihood of acquiring the disease or condition, and exposomic markers, relating to one or more environment factors over a defined or undefined period of time associated with having the disease or condition or likelihood of acquiring the disease or condition, and a combination thereof.

[0054] A “user” or “individual” as used herein may refer to, but is not limited to, an individual or group of individuals exploiting, accessing or using an embodiment of the invention. This includes, private individuals, employees of organizations and / or enterprises, members of community organizations, members of charity organizations, men, women and children. In its broadest sense the user may further include, but not be limited to, mechanical systems, robotic systems, android systems, etc. that may be characterised by an ability to exploit one or more embodiments of the invention.

[0055] A “server” as used herein, and throughout this disclosure, refers to one or more physical computers co-located and / or geographically distributed running one or more services as a host to users of other computers, PEDs, FEDs, etc. to serve the client needs of these other users. This includes, but is not limited to, a database server, file server, mail server, print server, web server, gaming server, or virtual environment server.

[0056] An "application" (commonly referred to as an "app") as used herein may refer to, but is not limited to, a "software application", an element of a "software suite", a computer program designed to allow an individual or user to perform an activity, a computer program designed to allow an electronic device to perform an activity, and a computer program designed to communicate with electronic devices which may be local, fixed, portable remote. Generally, within the following description with respect to embodiments of the invention an application is generally presented in respect of software permanently and / or temporarily installed upon a server or electronic device.

[0057] "Biometric" information as used herein may refer to, but is not limited to, data relating to a user characterised by data relating to a subset of conditions including, but not limited to, their environment, medical condition, biological condition, physiological condition, chemical condition, ambient environment condition, position condition, neurological condition, drug condition, and one or more specific aspects of one or more of these said conditions. Accordingly, such biometric information may include, but not be limited, blood oxygenation, blood pressure, blood flow rate, heart rate, temperate, fluidic pH, viscosity, particulate content, solids content, altitude, vibration, motion, perspiration, EEG, ECG, energy level, etc. In addition, biometric information may include data relating to physiological characteristics related to the shape and / or condition of the body wherein examples may include, but are not limited to, fingerprint, facial geometry, baldness, DNA, hand geometry, odour, and scent. Biometric information may also include data relating to behavioral characteristics, including but not limited to, typing rhythm, gait, and voice.

[0058] "User information" as used herein may refer to, but is not limited to, user behavior information and / or user profile information acquired from one or more electronic devices. It may also include a user's biometric information, an estimation of the user's biometric information, or aprojection / prediction of a user's biometric information derived from current and / or historical biometric information.

[0059] A “sensor” as used herein may refer to, but is not limited to, a transducer providing an electrical output push environmental sensors, medical sensors, biological sensors, chemical sensors, ambient environment sensors, position sensors, motion sensors, thermal sensors, infrared sensors, visible sensors, RFID sensors, and medical testing and diagnosis devices.

[0060] A "wearable device" or "wearable sensor" relates to miniature electronic devices that are worn by the user including those under, within, with or on top of clothing including electronic devices and / or sensors attached to or coupled to the user’s body and are part of a broader general class of wearable technology which includes "wearable computers" which in contrast are directed to general or special purpose information technologies and media development. Such wearable devices and / or wearable sensors may include, but not be limited to, smartphones, smart watches, e-textiles, smart shirts, activity trackers, smart glasses, environmental sensors, medical sensors, biological sensors, physiological sensors, chemical sensors, ambient environment sensors, position sensors, neurological sensors, drug delivery systems, medical testing and diagnosis devices, and motion sensors.

[0061] "Electronic content" (also referred to as "content" or "digital content") as used herein may refer to, but is not limited to, any type of content that exists in the form of digital data as stored, transmitted, received and / or converted wherein one or more of these steps may be analog although generally these steps will be digital.

[0062] A "profile" as used herein, and throughout this disclosure, refers to a computer and / or microprocessor readable data file comprising data relating to a user or individual. Such profiles may be established by a manufacturer / supplier / provider of an electronic device, a sensor, service, etc. or they may be established by the user through a user interface for an electronic device, a sensor, a service or another electronic device in communication with other devices, sensors, servers or a service provider, third-party provider, an enterprise, a government organization and the like.

[0063] A "computer file" (commonly known as a file) as used herein, and throughout this disclosure, refers to a computer resource for recording data discretely in a computer storage device, this data being electronic content. A file can be opened, read, modified, copied, and closed with one or more software applications an arbitrary number of times. Typically, files are organized in a file system which can be used on numerous different types of storage device exploiting different kinds of media which keeps track of where the files are located on the storage device(s) and enables user access.

[0064] "Metadata" as used herein, and throughout this disclosure, refers to information stored as data that provides information about other data. Many distinct types of metadata exist, including but not limited to, descriptive metadata, structural metadata, administrative metadata, reference metadata and statistical metadata. Descriptive metadata may describe a resource for purposes such as discovery and identification and may include, but not be limited to, elements such as title, abstract, author, and keywords. Structural metadata relates to containers of data and indicates how compound objects are assembled and may include, but not be limited to, how pages are ordered to form chapters, and typically describes the types, versions, relationships and other characteristics of digital materials. Administrative metadata may provide information employed in managing a resource and may include, but not be limited to, when and how it was created, file type, technical information, and who can access it. Reference metadata may describe the contents and quality of statistical data whereas statistical metadata may also describe processes that collect, process, or produce statistical data. Statistical metadata may also be referred to as process data.

[0065] An "artificial intelligence system" (AIS) as used herein, and throughout disclosure, refers to an electronic system, either localized or distributed, providing machine intelligence or machine learning. An AIS may refer to analytical, human inspired, or humanized artificial intelligence. An AIS may refer to the use of one or more machine learning algorithms and / or processes (MLAPs). An AIS may employ, for example, one or more of an artificial network, decision trees, support vector machines, Bayesian networks, and one or more genetic algorithms. An AIS may employ a training model or federated learning.

[0066] "Machine Learning Algorithms and / or Processes" (MLAPs) or more specifically machine learning processes as used herein refers to, but is not limited, to programs, algorithms or softwaretools, which allow a given electronic device or software program to learn and adapt its functionality based on information processed by it or by other independent processes. These MLAPs are in practice, gathered from the result of said process which produce data and or algorithms that lend themselves to prediction. This prediction process allows MLAP-capable devices to behave according to guidelines initially established within its own programming but which evolve as a result of the MLAPs they execute. MLAPs as employed by an AIS may include, but not be limited to, supervised learning, unsupervised learning, cluster analysis, reinforcement learning, feature learning, sparse dictionary learning, anomaly detection, association rule learning, and inductive logic programming.

[0067] Figure 1 A.1 and 1 A.2 show the biomarkers screened by the trained computer model. These include amino acids, coagulations factors with immunomodulatory functions, liver function and hepatic markers, coagulation and hemostatis biomarkers, metabolic markers involved in carbohydrate metabolism, lipid metabolism, nucleotide and nucleic acid metabolism, amino acid metabolism and other metabolic markers, cardiovascular and lipid metabolism, structural and extracellular matrix proteins, enzymes and enzyme inhibitors, hormones and growth factor binding proteins, renal function, neutrotransmitter metabolism, hematological markers, lipids, including lipoproteins such as apolipoproteins, sphingolipids, lysophospholipids, glycerosphingolipids, acylcarnitines, including medium and long chain acylcarnitines.

[0068] The machine learning model designed by the inventors was trained on large metabolomic and proteomic datasets and assigned three distinct endotypes to subject samples (Table 1 and Figures 2-5). The inventors used the information to design a computer-implemented method to assign a sample to at least one of the three endotype classes that cause disease progression when at least 20%, 30%, 40%, 50%, 60% or more of the biomarkers in at least one of the endotype classes is out of a normal range.Computer-implemented endotype assignment and scoring

[0069] Figure IB shows endo-typing of subjects into one or more of the three endotyping categories. Endotype 1 is characterized by dyslipidemia (e.g., sphingolipid dysregulation); (ii) endotype 2 is characterized by metabolic aberrations (e.g., amino acid and metabolism dysregulation); and (iii) endotype 3 is characterized by inflammation and oxidative stress. Theendotypes are assigned based on a computational model using the multimodal batch processing that has been trained with proteomic and metabolic information obtained from a plurality of samples from at least a subset of subjects having Alzheimer’s disease.

[0070] Figure 1C indicates how the endotyping data is used to assign a subject’s sample in one of the three endotypes identified by the computer model based on determining whether biomarkers in each endotype are out of range. A sample associated with a subject is assigned to a dyslipidemia endotype 1 if at least 20% of the sphingolipids tested are out of range (8 out of 9 in this example). Likewise, a sample associated with a subject is assigned to endotype 2 characterized by metabolic aberrations (e.g., amino acids) if at least 20% are out of range, in this case 12 out of the 19 amino acids are out of a normal range. A subject sample is assigned to endotype 3 characterized by inflammation and oxidative stress if at least 20% of the inflammatory markers tested are out of range, which is 10 out of 14 in this example. It will be appreciated that a subject sample may be assigned 1, 2 or all 3 of the endotypes.

[0071] Figure ID shows how data 5 from a subject sample is stratified to assign an endotype to the sample. Markers for each endotype category 1-3 are assigned numerical scores by the computerized method. For example, a score of 1 assigned by the computerized method indicates that the marker in a given endotype is higher than normal (out of range), while 0 is normal, and a negative number is lower than normal (out of range). In the example, it can be seen that in the metabolic aberrations endotype 2, for one of the samples analyzed by the computer model, glutamine is out of range, while other markers tested, namely alanine, arginine, asparagine, aspartic acid, glutamic acid, glycine, proline, serine and tyrosine are not out of range (see Figure ID magnified region 4 from dataset). Other subject samples, however, would be assigned this endotype (samples with blocks of darkened regions assigned a score that is positive or negative, meaning out of range).

[0072] Within other embodiments of the invention a score (e.g., numerical or any other scoring metric) may be assigned to a marker for an individual in dependence upon a magnitude of the marker measured from a sample for that individual or a magnitude of a difference from a nominalaverage of the marker measured from the sample for that individual where positive deviations may be larger than negative ones or vice-versa etc. according to the nominal average. The score may, therefore rather than being set to a value of a set of specific values, e.g. - 1, 0, _1, may be normalized between -1 and 1, or it may be normalized to a different range for the biomarker or normalized relative to a magnitude of another marker of the individual or normalized to an average or a weighted average of a set of other biomarkers of the individual for example. Optionally, it may be normalized based upon a deviation from a band defining a normal or nominal range, such that for example a biomarker value of 30 is scored based upon it being -10 relative to a nominal band of 40-60 and a biomarker value of 80 is scored based upon it being +20 relative to the nominal band. Within embodiments of the invention the score may be linearly defined based upon a magnitude of the biomarker or non-linearly defined based upon the magnitude of the biomarker, e.g. logarithmic or based upon a defined mathematical function.

[0073] Within embodiments of the invention the nominal value, nominal band etc. upon which a score is established may be constant across all individuals or it may be constant for all individuals within a demographic to which the individual belongs which may be established in dependence upon a profile of a user and may include factors such as ethnicity, age, gender, nationality, parental nationality, residential location, blood type, etc. Within other embodiments of the invention the nominal value, nominal band etc. upon which a score is established may be established in dependence upon data acquired from one or more sensors associated with the individual.

[0074] Within other embodiments of the invention the nominal value, nominal band etc. upon which a score is established may be established in dependence upon data acquired from one or more sensors associated with the individual or an environment associated with the individual.

[0075] Within embodiments of the invention the score associated with a biomarker may be scaled in dependence upon one or more factors. Each factor of the one or more factors may be established in dependence upon a profile of the user such that the weighted score is established in dependence upon the one or factors for the individual rather than a score common to all individuals with the same biomarker value established from the testing. The profile of the individual may define for example the individual’s ethnicity, age, gender, nationality, parental nationality, residential location and blood type for example. The profile of the individual may also link to one or moreother profiles where each other profile is associated with another individual related to the individual, such as father, mother, brother, sister, son and daughter for example. The other profiles may define, for example, one or more diseases the another individual had or has, one or more conditions that the another individual had or has, and a cause of death or invalidity of the another individual.

[0076] Within embodiments of the invention the trained computational model establishes an endotype class of the number of endotype classes in dependence upon scores of at least a defined percentage of the plurality of biomarkers meeting a predetermined condition or a defined number of biomarkers of the plurality of biomarkers. Within embodiments of the invention the percentage of the plurality of biomarkers meeting a predetermined condition or the defined number of biomarkers of the plurality of biomarkers may be constant for all individuals or it may be constant for all individuals within a particular demographic or established in dependence upon one or more factors of the individual defined within a profile of the individual which may define for example the individual’s ethnicity, age, gender, nationality, parental nationality, residential location and blood type for example. The profile of the individual may also link to one or more other profiles where each other profile is associated with another individual related to the individual, such as father, mother, brother, sister, son and daughter for example. The other profiles may define, for example, one or more diseases the another individual had or has, one or more conditions that the another individual had or has, and a cause of death or invalidity of the another individual.

[0077] Within embodiments of the invention the training data set for training the computational model may be common to all individuals or it may be constant for all individuals within a particular demographic or established in dependence upon one or more factors of the individual defined within a profile of the individual such that the trained computational model is established for each individual rather than for all individuals or all individuals within a particular demographic. The profile of the individual may define one or more factors that are employed to filter a data set to establish the training data set for training the computational model. The one or more factors may include ethnicity, age, gender, nationality, parental nationality, residential location and blood type for example. The profile of the individual may also link to one or more other profiles where each other profile is associated with another individual related to the individual, such as father, mother, brother, sister, son and daughter for example. The other profiles may define, for example, one ormore diseases the another individual had or has, one or more conditions that the another individual had or has, and a cause of death or invalidity of the another individual.

[0078] Within embodiments of the invention the training data set for training the computational model may be filtered in dependence upon one or more filter factors. Within an embodiment of the invention a filter factor of the one or more filter factors may be established in dependence upon the endotype class for which the computational model will be trained using the filtered training data. Within an embodiment of the invention a filter factor of the one or more filter factors may be a defined date range, e.g. a defined period of time prior to the computational model being trained, a defined period of time prior to onset of a disease or condition associated with the endotype class for which the computational model will be trained and a time frame where the time frame is established in dependence upon one of a value of a biomarker of the plurality of markers of the individual, a statistical calculation for a biomarker of the plurality of markers of the individual from multiple samples acquired from the individual prior to training the computational model and a statistical calculation for a subset of the plurality of markers of the individual from multiple samples acquired from the individual prior to training the computational model.

[0079] Within embodiments of the invention the trained computational model establishes an endotype class of the number of endotype classes where the trained computational model is training using a data set as outlined above. Within embodiments of the invention the trained computational model may be common to all endotype classes of the plurality of endotype classes. Within other embodiments of the invention the trained computational model may be one of a set of trained computational models where each trained computational model of the set of trained computational models is for a defined subset of the plurality of endotype classes. Within another embodiment of the invention each trained computational model of the set of trained computational models is specific to a single endotype class.Improved computer function relative to prior computer-implemented AD dataset analysis

[0080] Various studies have assigned AD progression to multiple deregulated pathways such as neurotransmitter synapses, oxidative stress, inflammation, vitamins, complement and coagulation pathways, cell death, cellular senescence, energy metabolism, genomic integrity, glia, immunesystem, metal ion homeostasis, oxidative stress, proteostasis, and synaptic function (Kodam et al., 2023, Nature, 13:3695; and Shokhirev and Johnson, 2022, Ageing Research Reviews, 81:101721). Kodam et al., (ibid) identified 2 common pathways between transcriptomics and metabolomics data, 3 common pathways between proteomics and metabolomics and 13 common pathways between transcriptomics and proteomics.

[0081] While this information proves useful for understanding the various complicated pathways that contribute to the disease, the inventors have found that assigning a large number of endotypes to each of these pathways by machine learning may be impractical and results in increased data center usage, which has become an ongoing problem in clinical settings as new digital technologies are adopted.

[0082] To the inventors’ knowledge no patient portal has been developed that simplifies disease progression into three endotypes, with treatment schedules for each endotype, along with biomarker tracking.

[0083] In addition, the inventors have found that improved computing efficiency by avoiding the assessment of multiple, complicated pathways as per prior methods, reduces data center usage or energy efficiency. Such surprising reductions are set forth below.

[0084] In some embodiments, the method for endotyping Alzheimer's disease (AD) using a machine learning model can be significantly enhanced by batching multiple disease models and endotypes together. This approach, referred to herein as "multi-modal batch processing," allows for the simultaneous analysis of various biomarkers and disease pathways, leading to a more efficient and less resource-intensive computational process. The discussion below outlines how the present disclosure can leverage this method to reduce AI runtime and energy consumption.

[0085] Traditional methods for analyzing disease progression often focus on single pathways, such as inflammation, metabolic dysregulation or sphingolipid dysregulation. These methods, while informative, can be computationally demanding, especially when analyzing the vast biomarker data generated by high-throughput screening. Relying on a large number of complex pathways for analysis also increases data center usage, an ongoing problem in clinical settings with the adoption of new digital technologies. The computational power required for these models can leadto significant energy consumption from running and cooling servers, contributing to greenhouse gas emissions.

[0086] The present machine learning model simplifies the complex task of AD risk assessment by identifying three distinct endotypes:

[0087] Sphingolipid Dysregulation: Characterized by abnormal levels of sphingolipids, such as ceramides and sphingomyelins.

[0088] Amino Acid and Metabolism Dysregulation: Defined by deviations in amino acid levels, such as glutamate, glycine, and GABA, and other metabolic biomarkers.

[0089] Inflammation and Oxidative Stress: Marked by elevated levels of pro-inflammatory cytokines, chemokines, and immune-related proteins.

[0090] Instead of running separate models for each disease pathway, multi-modal batch processing involves training a single, consolidated model on the combined proteomic and metabolomic datasets. This single model is designed to simultaneously identify and classify all three endotypes. By processing a batch of patient samples at once, the model can leverage the shared computational resources and avoid the redundant calculations that would be necessary for running three separate models.

[0091] The flowsheet of Figure 5 illustrates the computerized method utilizing multi-modal batch processing. Proteomic and metabolomic datasets 20 from normal and Alzheimer’s patients are used to train a computational model using multi-modal batch processing 30, thereby producing a trained computational model 40. The trained computational model is then used to measure proteomic and metabolomic biomarkers from a sample of a human subject via a high throughput spectroscopic assay 50. The measured biomarkers are categorized into at least one of three endotype classes by the trained computational model 60. Based on the assigned endotype, a treatment is displayed on a graphical user interface 70.

[0092] Using multi-modal batch processing, the inventors estimate a significant reduction in AI runtime and, consequently, computational energy use. By consolidating the analysis of threeendotypes into a single, batched process, the overall computational workload is reduced. Instead of the sum of the runtimes for each individual model, the batched model can be processed in a single, streamlined operation.

[0093] Assuming that each individual model takes approximately the same amount of time to run, consolidating them into a single batched process could theoretically reduce the total Al runtime by up to 30%. This is because a single, optimized model would avoid the overhead of loading and running separate processes. The reduction in Al runtime directly translates to a decrease in the demand for computational resources, such as servers and data centers. A lower computational load means less energy is needed to power and cool the servers, which in turn reduces operational costs and the overall environmental impact. This approach allows for a comprehensive and accurate risk assessment without the excessive energy usage that has become a problem in clinical settings. Advantages of the consolidated approach

[0094] Improved Computer Function: The present method is simpler to implement than previous approaches that rely on identifying multiple, complex disease etiologies. The simplified training model improves computer function without compromising predictive power. The consolidated model would maintain this simplicity while offering even greater efficiency.

[0095] Reduced Data Center Usage: By avoiding the assessment of multiple, complicated pathways, the method disclosed herein reduces the energy consumption and data center usage associated with conventional methods. The batching process would amplify this benefit by processing more data in a shorter amount of time.

[0096] Enhanced Predictive Accuracy: The ability to identify three distinct endotypes avoids false negatives that might occur with methods that focus on a single disease mechanism. Training of the proteomic and metabolomic data with multi-modal batch processing steps provides the added advantage of integrating all three endotypes into a single, powerful predictive model. Training of the data sets using multi-modal batch processing represents a significant step forward to provide a practical, efficient, and environmentally conscious method for assessing AD progression and risk.

[0097] In some exemplary embodiments, assigning additional endotype classes beyond 5 increases data center usage relative to the instant invention in which 3 distinct endotype classes are assigned to a dataset.

[0098] In some embodiments, the training is carried out by multimodal batch processing in which each batch comprises data elements that are synchronized across the modalities for a given set of samples taken from Alzheimer’s patients. Unlike conventional mini -batching that handles a single homogeneous data type, multimodal batch processing maintains cross-modality alignment and parameter updating during training or inference of a computation model. In certain embodiments, the multimodal batch processing is implemented in a shared computational graph that allows modality specific feature tensors to be loaded, normalized and encoded in parallel followed by join parameter updates in one or more shared network layers. This approach reduces computational latency and memory transfer overhead associated with sequential, single modality integration, and can improve model convergence stability and prediction accuracy in proteomic and metabolic analysis. In some embodiments, multimodal batch processing utilizes distributed or GPU-based parallelization to optimize throughput and hardware utilization.Dataset for endotyping

[0099] In certain embodiments, an initial step of the computerized method includes measurement of the concentration levels of markers within each endotype class by an instrument, which may include a mass spectrometry unit, including but not limited to gas chromatography mass spectrometry (GC-MS) GC and liquid chromatography mass spectrometry (e.g., LC-MS, LC-MS-MS, LC-MRM, LC-SIM, and LC-SRM) unit. In some embodiments, the markers in each class are measured by a spectroscopic unit, the spectroscopic unit being selected from the group consisting of liquid chromatography, gas chromatography, liquid chromatography mass spectrometry, gas chromatography mass spectrometry, high performance liquid chromatography mass spectrometry, capillary electrophoresis mass spectrometry, nuclear magnetic resonance spectrometry (NMR), raman spectroscopy, and infrared spectroscopy.

[0100] In some embodiments, the mass spectrometry process for determining whether the markers in each endotype are out of a normal range comprises enzymatic or chemical digestion ofthe proteins or peptide fragments thereof of a sample obtained from a subject into peptide fragments. The peptide fragments are optionally separated and / or ionized and captured by mass spectrometry. The digestion may comprise a proteolytic digestion involving treating a preparation comprising the Alzheimer’ s-related proteins with an acid, base, or an enzyme such as trypsin or other proteolytic enzyme. One embodiment comprises a shotgun proteomics quantification in which the whole proteins in a complex mixture, such as serum, urine, and cell lysates, are hydrolyzed or otherwise cut into peptides and followed by multidimensional HPLC-MS, which aims to generate a global profile of protein mixtures as genome “shotgun” sequencing.Endotype assignment of a sample

[0101] In some embodiments, the biomarkers are assigned a numerical score by the computational model corresponding to whether the marker is out of range or within range. For example, a biomarker within range can be assigned 0 and a biomarker out of range can be assigned a value that is positive or negative depending on whether the biomarker is elevated relative to a normal range.

[0102] In certain embodiments, a biomarker is out of range if a biomarker in a sample dataset differs by about 20% or more, about 30% or more, about 40% or more, about 50% or more, about 60% or more, or about 70% or more relative to a normal concentration range of a biomarker.

[0103] Examples of sphingolipid biomarkers measured for assignment of endotype 1 include hydroxysphingomyelines, C14:l, C16:l, C22:l, C22:2 and C24:l and sphingomyelines C16:0, C16:l, C18:0, C18:l and C20:2 (Figure 2A and 2B). If greater than 20%, 30%, 40%, 50%, 60% or 80% of the sphingolipid biomarkers are out of range, then the subject dataset is assigned to endotype 1 characterized by sphingolipid dysregulation. In some embodiments, endotype 1 is assigned to a sample dataset ifgreater than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 or 15 sphingolipids are out of range.

[0104] In addition to sphingolipids, certain lysophospholipids may be out of range as well and include lysphophatidylcholine acyl C16:0, 16:1, 17:0, 18:0, 18:2, 20:4, 24:0, 26:0, 26:1, 26:0 and 28:1. In some examples, if the biomarker data has sphingolipid dysregulation, and greater than 20%, 30%, 40%, 50%, 60% or 80% of the lysphophatidylcholine biomarkers are out of range, then the subject dataset is assigned to endotype 1 characterized lipid dysregulation.

[0105] Examples of amino acid biomarkers measured for assignment of endotype 2 include essential and non-essential amino acids. Essential amino acids included histidine, isoleucine, leucine, lysine, methionine, methionine, phenylalanine, threonine, tryptophan and valine. Non-essential amino acids include alanine, arginine, asparagine, aspartic acid, glutamic acid, glutamine, glycine, proline, serine and tyrosine. Other amino acids and derivatives include alpha-aminoadipic acid, citrulline, gamma-amino butyric acid, homocysteine, methylhistidine, ornithine, taurine and trans-OH-proline. Other amino acids and derivatives that may be out of range include amino acids or derivatives thereof selected from alpha-aminoadipic acid, citrulline, gamma-aminobutyric acid, homocysteine, methylhistidine, ornithine, taurine and trans-OH-proline. If greater than 20%, 30%, 40%, 50%, 60% or 80% of the amino acid biomarkers are out of range, then the subject dataset is assigned to endotype 2 characterized by amino acid and metabolism dysregulation. In some embodiments, endotype 2 is assigned to a sample dataset if greater than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 or 15 amino acids are out of range.

[0106] Examples of inflammation and oxidative stress biomarkers measured for assignment of a sample to endotype 3 include complement proteins selected from C4b-binding protein alpha chain, complement Clq subcomponent subunit B, complement Clr subcomponent, complement Clr subcomponent-like protein, complement Cis subcomponent, complement Cis subcomponent, complement C2, complement C3, complement C4-B, complement C5, complement component C6, complement component C7, complement component C8 alpha chain, complement component C8 beta chain, complement component C9, complement factor B, complement factor D, complement factor H, complement factor I. Other inflammation and oxidative stress biomarkers included ficolin-3 and mannose-binding protein C. Additional inflammation and oxidative stress biomarkers identified by the model included trimethylamine N-oxide, putrescine, fibronectin, thrombospondin- 1. If greater than 20%, 30%, 40%, 50%, 60% or 80% of the inflammation and oxidative stress biomarkers are out of range, then the subject dataset is assigned to endotype 3 characterized by inflammation and oxidative stress. In some embodiments, endotype 3 is assigned to a sample dataset if greater than 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 or 15 inflammation and oxidative stress biomarkers are out of range.Optional treatment of each endotype

[0107] Examples of treatments for endotype 1 include sphingolipid-modulating drugs and dietary treatments aimed at restoring sphingomyelin levels to within a normal range.

[0108] In one embodiment, the drug used to adjust levels of a sphingolipid is a sphingomyelinase or cerimidase inhibitor. In one embodiment, the enzyme targeted to modulate sphingolipids is an nSMase 2, aSMase / ACDase, aCDase or a nDCDase / alkCDase.

[0109] Treatments for endotype 2 comprise adjusting amino acid levels and improving mitochondrial function. Potential treatments include amino acid supplements, metabolic enhancers, and dietary treatments for adjusting amino acid levels or other biomarkers within endotype 2 to within a normal range.

[0110] Treatment for endotype 3 comprises reducing inflammation and modulating immune responses. Examples of treatments include anti-inflammatory drugs and immunomodulatory therapies. Examples of drugs include non-steroidal anti-inflammatory drugs. Examples include p38 inhibitors.

[0111] In some embodiments, progression or regression of mild cognitive impairment or Alzheimer’s disease is monitored by obtaining a second metabolomic and proteomic dataset from the subject at a second time point and assigning the sample to one or more of the three endotypes. In some embodiments, progression or regression of mild cognitive impairment is monitored by obtaining a second metabolomic and proteomic dataset from the subject at a second time point and determining whether one or more biomarkers in the dataset previously out of range are within a normal range or outside of the normal range and, if one or more of the biomarkers are out of range, treating the subject so that the biomarkers fall within a normal range. In some embodiments, a patient portal is used to track progression or regression of mild cognitive impairment or Alzheimer’s disease. In some embodiments, longitudinal tracking icons are displayed on the patient portal to easily track changes in biomarker concentrations over time.

[0112] In some embodiments, at a subsequent time point after an initial dataset is obtained, the treatment results in greater than 20%, 30%, 40%, 50%, 60% or 80% of the sphingolipid biomarkers in endotype 1 assigned earlier as being out of range, falling within a normal range. In some embodiments, the subject is treated so that less than 1, 2, 3, 4 or 5 sphingolipids are out of range.In some embodiments, the subject is treated so less than 2 sphingolipids are out of range. In further embodiments, the subject is treated so that the subject dataset is no longer assigned to endotype 1.

[0113] In some embodiments, at a subsequent time point after an initial dataset is obtained, the treatment results in greater than 20%, 30%, 40%, 50%, 60% or 80% of the amino acid biomarkers in endotype 2 assigned earlier as being out of range, falling within a normal range. In some embodiments, the subject is treated so that less than 1, 2, 3, 4 or 5 amino acids are out of range. In some embodiments, the subject is treated so less than 2 amino acids are out of range. In further embodiments, the subject is treated so that the subject dataset is no longer assigned to endotype 2.

[0114] In some embodiments, at a subsequent time point after an initial dataset is obtained, the treatment results in greater than 20%, 30%, 40%, 50%, 60% or 80% of the inflammation and immune biomarkers in endotype 3 assigned earlier as being out of range, falling within a normal range. In some embodiments, the subject is treated so that less than 1, 2, 3, 4 or 5 inflammation and immune biomarkers in endotype 3 are out of range. In some embodiments, the subject is treated so less than 2 amino acids are out of range. In further embodiments, the subject is treated so that the subject dataset is no longer assigned to endotype 2.EXAMPLESExample 1: Machine learning of novel dataset identifies three AD endotypes

[0115] A comprehensive metabolomic and proteomic analysis of blood samples was conducted from patients with MCI and AD symptoms and / or diagnoses. Utilizing high-throughput spectroscopy, namely LC-MS (mass spec), the inventors quantified a wide array of metabolites and proteins, identifying patterns and abnormalities indicative of disease progression. Statistical analyses and a machine learning computerized model using supervised classification and normalization were employed to classify patients into distinct endotypes based on their metabolic and proteomic profiles.

[0116] The analysis by the computerized model identified a subset of patients whose risk of developing AD was primarily driven by abnormalities in sphingolipid levels. The results from samples tested by the computational model are shown in Table 1 and the endotyping data generated by the computer model is shown in Figures 2 and 3. Lipids identified by the model as beingelevated (1) above the normal range (0) as assigned by the computer model include hydroxysphingomyelines, C14:l, C16:l, C22:l, C22:2 and C24:l and sphingomyelines C16:0, C16:l, C18:0, C18:l and C20:2 (Figure 2A and 2B). Certain lysophospholipids were out of range as well and include lysphophatidylcholine acyl C16:0, 16:1, 17:0, 18:0, 18:2, 20:4, 24:0, 26:0, 26:1, 26:0 and 28:1. However, significantly more sphingolipids, specifically sphingomyelines C16:0, C16:l, C18:0, C18:l and C20:2, were identified as out of range by the computer model than the lysophophospholipids.

[0117] Patients in the sphingolipid dysregulation endotype exhibited significantly altered levels of specific sphingolipids, such as ceramides (hydroxysphingomyelines) and sphingomyelines C16:0, C16:l, C18:0, C18:l and C20:2. These changes were associated with increased neuroinflammation, neuronal apoptosis and impaired synaptic function. Surprisingly, other biomarkers in these patients, including amino acids and immune markers, remained within normal ranges (see Table 1 below), suggesting that sphingolipid dysregulation was the primary driver of their disease progression.Table 1: Endotyping of samples as identified by trained computational modelGlyceroAmino phosphoTestID - Amine Oxide - Sugar Acid ~ lipid - Sphingolipid - 1695631208 -0.5769230769 -0.6 -0.1 2665929747 -0.5192307692 -0.1 0 6546993780-0.4423076923 -0.1 0.25 6892993038 0 -0.5 -0.5576923077 0 0.4 7625441089 0 0 -0.4807692308 -0.2 0.05 5040715673 0 0 -0.3076923077 0.35 0.3 1701301622 0 0 -0.2115384615 0.05 0.25 5559680573 0 0 -0.1730769231 0 0.2 7920878274 0 0 -0.1730769231 -0.1 -0.15 1432866877 0 0 0.1730769231 0 0.45 9203384079 0 0 -0.1346153846 0 0.2 3265091508 0 0 -0.01923076923 0.85 0.85 6534171365 0 0 -0.01923076923 0.3 0.9 2803837018 0 0 0.02 0.5 0.9 5201967667 0 0 0.07692307692 0.6 0.55 9408769611 0 0 0.07692307692 0.75 0.9 6523318795 0 0 0.16 0.6 0.85 7533281411 0 0 0.09615384615 0.75 0.8 5738839694 0 0.5 0.01923076923 0.45 0.85 2585697226 0 0.5 0.03846153846 0.15 0.35 3131488433 0 1 0 0.05 0 1845192903 0 1 0.12 0.1 0.4 2619906783 0 1 0.1346153846 0.8 0.9 4796293944 0 1 0.5769230769 0.65 0.7 4759894233 1 1 0.02 0.15 0.35 1245879182 1 1 0.03846153846 0.5 0.5 4742637392 1 0 0.01923076923 0.05 0.35 4887474569 1 0 0.05769230769 0.05 0.3 1166082634 1 0 0.07692307692 0.1 0.3 3010941130 1 0 0.12 0.05 0.25 7625612713 1 0 0.01923076923 0.1 0.3

[0118] Patients in the amino acid and metabolism dysregulation endotype displayed significant deviations in the levels of specific amino acids, such as glutamate, glycine, and GABA and were assigned by the computational model to endotype 2 (Figure 2A-C and Figure 3). These abnormalities were accompanied by alterations in metabolic biomarkers, including acylcarnitinesand tricarboxylic acid (TCA) cycle intermediates. The dysregulation of amino acid metabolism was associated with oxidative stress, mitochondrial dysfunction, and impaired neurotransmission, contributing to AD progression.

[0119] Amino acids identified by the model for assignment to type 2 endotype included essential and non-essential amino acids. Essential amino acids included histidine, isoleucine, leucine, lysine, methionine, methionine, phenylalanine, threonine, tryptophan and valine. Non-essential amino acids include alanine, arginine, asparagine, aspartic acid, glutamic acid, glutamine, glycine, proline, serine and tyrosine. Other amino acids and derivatives include alpha-aminoadipic acid, citrulline, gamma-amino butyric acid, homocysteine, methylhistidine, ornithine, taurine and trans-OH-proline (Figures 2B and 2C). Other amino acids and derivatives that were out of range include amino acids or derivatives thereof selected from alpha-aminoadipic acid, citrulline, gamma-aminobutyric acid, homocysteine, methylhistidine, ornithine, taurine and trans-OH-proline (Figure 2C).

[0120] The amino acid and metabolism dysregulation endotype 2 uncovered by the computer model highlights the importance of metabolic health in neurodegenerative diseases. The inventors identified that treatments displayed on the computerized portal for this endotype would include instructions for modulating amino acid levels and / or improving mitochondrial function. Specific treatments displayed in the portal include amino acid supplements, metabolic enhancers, and dietary interventions designed to optimize metabolic health.

[0121] Patients in the inflammation and immune dysregulation endotype exhibited significantly higher levels of pro-inflammatory cytokines, chemokines, and immune-related proteins, such as C-reactive protein (CRP), Lipopolysaccharide-Binding Protein (LBP), and complement factors (Figure 2A, 2B and 2C). These patients also showed increased markers of immune activation, including elevated levels of complement proteins and immunoglobulins. The chronic inflammatory state in these patients was linked to neuronal damage, synaptic dysfunction, and accelerated cognitive decline.

[0122] Biomarkers associated with the inflammation and immune dysregulation endotype identified by the model included complement proteins selected from C4b-binding protein alpha chain, complement Clq subcomponent subunit B, complement Clr subcomponent, complementClr subcomponent-like protein, complement Cis subcomponent, complement Cis subcomponent, complement C2, complement C3, complement C4-B, complement C5, complement component C6, complement component C7, complement component C8 alpha chain, complement component C8 beta chain, complement component C9, complement factor B, complement factor D, complement factor H, complement factor I. Other inflammation and oxidative stress biomarkers included ficolin-3 and mannose-binding protein C (Figure 2B). Additional inflammation and oxidative stress biomarkers identified by the model included trimethylamine N-oxide, putrescine, fibronectin, thrombospondin- 1 (Figure 2C).

[0123] The inflammation and immune dysregulation endotype emphasizes the role of immune system health in AD progression. The inventors identified that treatments displayed on the computerized portal for this endotype would include instructions for reducing inflammation and modulating immune responses. Specific treatments displayed in the portal include antiinflammatory drugs, immunomodulatory therapies, and lifestyle changes aimed at reducing systemic inflammation.Example 2: Assignment of three distinct AD endotypes to datasets surprisingly reduces false negatives

[0124] Certain known AD risk assessments have focused on the importance of single pathways responsible for progression from mild cognitive impairment to AD. However, the inventors have discovered that overly simplistic approaches focusing on one pathway to assess AD risk may lead to false negatives.

[0125] The data above highlights this surprising observation and underscores the importance of assigning three distinct endotypes.

[0126] As shown in Table 1 above, sphingolipid dysregulation is identified as the primary driver of disease progression in the dataset and the sample is thereby assigned endotype 1. However, if an endotyping method in which only two categories were employed, e.g., endotypes 2 and 3 based on amino acid and metabolism dysregulation and inflammation and oxidative stress respectively, without sphingolipid endotyping (endotype 1), the patient sample would be assigned as “normal”, despite having a risk for developing AD. In other words, assignment into three endotypes may avoid false negatives and thereby assign a treatment to a patient that otherwise would not be treatedto slow AD progression. Accordingly, the method of the disclosure significantly improves patient outcomes relative to prior methods.

[0127] Moreover, the assignment of the three distinct endotypes described above by machine learning was itself a surprising observation. One approach in the literature for AD multi-omics analysis applying chemometric analysis of proteomic and metabolomic data relied on identifying specific biomarkers predictive of disease progression rather than assigning broad categories of endotypes to assign personalized treatments (Francois et al., 2022, Metabolites, 12:949). In addition, Francois et al. (ibid) identified glycerophospholipid ethanolamines as key drivers of disease, while the method of the present disclosure assigned sphingolipid dysregulation to a distinct endotype. In another study by Iturria-Medina et al., 2022, Neuroscience, Sci. Adv. 8, eabo6764 assigned three distinct AD endotypes from blood epigenomic, transcriptomomic, proteomic and metabolomic profiles. AD subtype 1 was characterized by metabolic alterations, and subtypes 2 and 3 by RNA and epigenetic alterations. Likewise, in this study, sphingolipids were not assigned a distinct endotype.Example 3: Novel patient portal for efficiently displaying MCI and / or AD progression health data to clinics, hospitals and patients, with respective treatment schedules

[0128] An example of a treatment display of the electronic portal is shown in Figure 4A. Icons are displayed in tabular format showing disease risk and / or biomarkers (left side, 7) identified by the model and treatment regimens for supplementing a subject’s diet (right side, 8) based on its findings. Clicking on a particular treatment regimen results in a box 9 displaying more treatment details for the subject. For example, as shown in Figure 4A, if Omega 3 fats in box 9 are part of the treatment, more information on the omega fats is displayed, e.g., instructions 10 for administering certain types of omega 3 fats. A similar window would be displayed for the Mediterranean diet (middle treatment regimen). The treatment regimen(s) of the electronic portal can be downloaded in csv format (top right) as well.

[0129] Figure 4B shows the patient portal interface 12 comprising a share sign-up link. This provides a link 13 with patients so they can create their own account to access the portal. The account will also be viewable by a hospital or clinic. There is also an option on the portal to createa profile 14 of the subject and to order a test. An icon is shown that when “clicked” automatically creates an email that is sent to the patient inviting them to create a profile.

[0130] Figure 4C shows longitudinal tracking of biomarkers over time 13 after a patient has been assigned an Alzheimer’s treatment schedule. The results are depicted pictorially as “high”, “medium” and “low” 14 in easy-to-follow longitudinal bars 15, 16, 17 and 18 corresponding to risk scores (derived from biomarker score or levels) reported in defined periods, e.g., every three months (4 times per year) as shown. In the example depicted, the risk score was 30 in October 2019, 41 in January, 2020, 66 in September, 2020 and 100 in October, 2020, with a higher score depicting a lowered risk of Alzheimer’s. In this case, the treatment resulted in a lowered risk of Alzheimer's over the tracking period and a message “Nice Work!” is displayed to the patient.

[0131] While particular embodiments of the present disclosure have been illustrated and described, it would be obvious to those skilled in the art that various other changes and modifications can be made without departing from the scope of the present disclosure. It is therefore intended to cover in the appended claims all such changes and modifications that are within the scope of this disclosure.

[0132] Whilst the embodiments of the invention have been described with respect to three endotypes it would be evident that the embodiments of the invention by exploiting trained computational models can be executed rapidly for an individual for larger numbers of endotypes without departing from the scope of the invention. Within embodiments of the invention the trained computational model processes a subset of multi-omics, including metabolomic, proteomic, and environmental exposures, data associated with the individual where the subset of multi-omics data comprises a first subset established in dependence upon processing one or more samples acquired from the individual and processed by an assay and a second subset established in dependence upon data acquired from one or more sensors associated with the individual.

[0133] Specific details are given in the above description to provide a thorough understanding of the embodiments. However, it is understood that the embodiments may be practiced without these specific details. For example, circuits may be shown in block diagrams in order not to obscure the embodiments in unnecessary detail. In other instances, well-known circuits, processes, algorithms,structures, and techniques may be shown without unnecessary detail or not shown in order to avoid obscuring the embodiments.

[0134] Implementation of the techniques, blocks, steps and means described above may be done in various ways. For example, these techniques, blocks, steps and means may be implemented in hardware, software, or a combination thereof. For a hardware implementation, the processing units may be implemented within one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, micro-controllers, microprocessors, other electronic units designed to perform the functions described above and / or a combination thereof.

[0135] Also, it is noted that the embodiments may be described as a process which is presented in a manner similar to a flowchart, a flow diagram or a data flow diagram. Although a flowchart may describe the operations as a sequential process, many of the operations can be performed in parallel or concurrently. In addition, the order of the operations may be rearranged. A process is terminated when its operations are completed, but could have additional steps not included in the figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc. When a process corresponds to a function, its termination corresponds to a return of the function to the calling function or the main function.

[0136] Furthermore, embodiments may be implemented by hardware, software, scripting languages, firmware, middleware, microcode, hardware description languages and / or any combination thereof. When implemented in software, firmware, middleware, scripting language and / or microcode, the program code or code segments to perform the necessary tasks may be stored in a machine readable medium, such as a storage medium. A code segment or machine-executable instruction may represent a procedure, a function, a subprogram, a program, a routine, a subroutine, a module, a software package, a script, a class, or any combination of instructions, data structures and / or program statements. A code segment may be coupled to another code segment or a hardware circuit by passing and / or receiving information, data, arguments, parameters and / or memory content. Information, arguments, parameters, data, etc. may be passed, forwarded, ortransmitted via any suitable means including memory sharing, message passing, token passing, network transmission, etc.

[0137] For a firmware and / or software implementation, the methodologies may be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. Any machine-readable medium tangibly embodying instructions may be used in implementing the methodologies described herein. For example, software codes may be stored in a memory. Memory may be implemented within the processor or external to the processor and may vary in implementation where the memory is employed in storing software codes for subsequent execution to that when the memory is employed in executing the software codes. As used herein the term “memory” refers to any type of long term, short term, volatile, nonvolatile, or other storage medium and is not to be limited to any particular type of memory or number of memories, or type of media upon which memory is stored.

[0138] Moreover, as disclosed herein, the term “machine-readable medium” may represent one or more devices for storing data, including read only memory (ROM), random access memory (RAM), magnetic RAM, core memory, magnetic disk storage mediums, optical storage mediums, flash memory devices and / or other machine readable mediums for storing information. The term “machine-readable medium” includes, but is not limited to portable or fixed storage devices, optical storage devices, and / or various other mediums capable of storing, containing or carrying instruction(s) and / or data. Whilst processed data or executing code may be stored within transitory storage media the underlying software code, which when executed controls one or more processes of a microprocessor or microcontroller, is stored within non-transitory storage media.

[0139] In alternative embodiments, a device (or system) may operate as a standalone device or may be connected, e.g., networked to other devices, in a networked deployment, the device may operate in the capacity of a server or a client machine in server-client network environment, or as a peer device in a peer-to-peer or distributed network environment. The device may be, for example, a computer, a server, a cluster of servers, a cluster of computers, a web appliance, a distributed computing environment, a cloud computing environment, or any device capable of executing a set of instructions (sequential or otherwise) that specify actions to be taken by that device. The term “device” may also be taken to include any collection of devices that individuallyor jointly execute a set (or multiple sets) of instructions to perform any one or more of the methodologies discussed herein.

[0140] The foregoing disclosure of the exemplary embodiments of the present invention has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many variations and modifications of the embodiments described herein will be apparent to one of ordinary skill in the art in light of the above disclosure. The scope of the invention is to be defined only by the claims appended hereto, and by their equivalents.

Claims

WE CLAIM:

1. A method to assess progression or regression of mild cognitive impairment or Alzheimer’s disease or a risk of developing Alzheimer’s disease based on proteomic and metabolomic data of one or more samples obtained from a human subject, the method comprising the steps of:a) measuring a plurality of at least 20 proteomic and metabolomic biomarkers from a sample of the human subject via a high-throughput spectroscopic assay, the plurality of biomarkers categorized into at least one of three endotype classes that cause disease progression, the endotypes selected from (i) endotype 1 characterized by sphingolipid dysregulation; (ii) endotype 2 characterized by amino acid and metabolism dysregulation; and (iii) endotype 3 characterized by inflammation and oxidative stress, the endotypes assigned based on a trained computational model that was trained with proteomic and metabolic datasets obtained from a plurality of samples from Alzheimer’s disease patients and normal individuals, wherein at least 3 biomarkers are measured in each of the three endotype classes;b) assigning the initial sample to at least one of the three endotype classes that cause disease progression when at least 20% of the biomarkers in at least one of the endotype classes is out of a normal range;c) optionally, via a graphical user interface, assigning a treatment to treat at least one of the three endotype classes and optionally treating the subject or causing the treating by administering a sphingolipid-modulating drug if the initial sample is assigned to endotype 1; one or more amino acid supplements and / or metabolic enhancers if the initial sample is assigned to endotype 2; and / or an anti-inflammatory drug if the initial sample is assigned to endotype 3;d) optionally obtaining a second sample from the subject at a second time point;e) optionally measuring a plurality of at least 20 proteomic and metabolomic biomarkers from the second sample via a high-throughput spectroscopic assay, wherein at least 5 biomarkers are measured in at least one of the three endotype classes;f) optionally determining whether there is an increase or decrease of the percentage of biomarkers in the assigned endotype class or classes of the second sample that are out of a normal range relative to the first sample and optionally assigning diseaseprogression if there is an increase in the biomarkers out of range or regression if there is a decrease in the biomarkers out of range; andg) optionally if a disease progression is assigned to the second sample, assigning a second treatment for at least one of the three endotype classes if disease progression is assigned to the second sample and optionally treating the subject by administering a sphingolipid-modulating drug if the second sample is assigned to endotype 1; amino acid supplements and / or a metabolic enhancer if the second sample is assigned to endotype 2; and / or an anti-inflammatory drug if the second sample is assigned to endotype 3.

2. The method of claim 1, wherein the initial sample is assigned to endotype 1 and wherein the at least 5 biomarkers are selected from a ceramide and / or a sphingomyelin.

3. The method of claim 2, wherein the sphingolipid is selected from a C14:l, C16:l, C22:l, C22:2 or C24:l hydroxysphingomyelin or a C16:0, C16:l, C18:0, C18:l and C20:2 sphingomyelin.

4. The method of claim 1, wherein the initial sample is assigned to endotype 2 and wherein the at least 3 biomarkers in endotype 2 that are measured are a metabolite, protein and / or amino acid.

5. The method of claim 4, wherein the at least 3 biomarkers measured in endotype 2 are essential amino acids selected from histidine, isoleucine, leucine, lysine, methionine, methionine, phenylalanine, threonine, tryptophan and valine; non-essential amino acids selected from alanine, arginine, asparagine, aspartic acid, glutamic acid, glutamine, glycine, proline, serine and tyrosine; or amino acids or derivatives thereof selected from alpha-aminoadipic acid, citrulline, gamma-aminobutyric acid, homocysteine, methylhistidine, ornithine, taurine and trans-OH-proline.

6. The method of claim 1, wherein the initial sample is assigned to endotype 3 and wherein the at least 3 biomarkers in endotype 3 that are measured are selected from pro-inflammatory cytokines, chemokines, and immune-related proteins, such as C-reactive protein (CRP), Lipopolysaccharide-Binding Protein (LBP), and complement factors.

7. The method of claim 6, wherein the at least 3 biomarkers measured in endotype 3 are selected from C4b-binding protein alpha chain, complement Clq subcomponent subunit B, complement Clr subcomponent, complement Clr subcomponent-like protein, complement Cis subcomponent, complement Cis subcomponent, complement C2, complement C3, complementC4-B, complement C5, complement component C6, complement component C7, complement component C8 alpha chain, complement component C8 beta chain, complement component C9, complement factor B, complement factor D, complement factor H, complement factor I, ficolin-3, mannose-binding protein C, trimethylamine N-oxide, putrescine, fibronectin and thrombospondin-1.

8. The method of claim 1, wherein the graphical user interface is a portal that is a patient portal or a clinician portal.

9. The method of claim 8, wherein the portal comprises a graphical icon depicting one or more risk scores for Alzheimer’s disease or for progression thereof.

10. The method of claim 9, wherein portal comprises graphical icons depicting biological pathways correlated with the three endotype classes.

11. The method of claim 10, wherein the graphical icons comprises a score or data correlated with the three endotype classes.

12. The method of claim 1, wherein the portal comprises a graphical icon depicting the treatment pathway and wherein the treatment is exportable as a set of data in computer readable format.

13. The method of claim 12, wherein the graphical icon when selected causes a window to be displayed with further treatment data.

14. The method of claim 12, wherein the set of data is in.csv or.pdf format.

15. The method of claim 13 or 14, wherein the treatment comprises supplement data.

16. The method of claim 1, wherein the portal comprises one or more graphical icons depicting comorbidity risks that are detected in the initial sample.

17. The method of claim 1, wherein the portal displays one or more biological pathways resulting from one or more of the biomarkers that are out of range.

18. The method of claim 1, wherein the assigning of the initial sample to the at least one of the three endotype classes that cause disease progression occurs when at least 30%, 40%, 50%, 60%, 70% or 80% of the biomarkers in at least one of the endotype classes is out of a normal range.

19. The method of claim 1, wherein at least 4, 5, 6, 7, 8, 9 or 10 biomarkers are measured in each of the three endotype classes.

20. The method of claim 1, wherein one or more undiagnosed comorbidity diseases or conditions are identified from the plurality of biomarkers measured in the initial sample.

21. The method of any one of claims 1 to 20, wherein at least 4, 5, 6, 7, 8, 9 or 10 biomarkers are measured in each endotype class.

22. The method of any one of claims 1 to 21, wherein the trained computational model has been trained by executing multimodal batch processing operations by concurrently processing batches of the proteomic and metabolomic data sets, and wherein the multimodal batch processing reduces computation latency relative to sequential processing.

23. A system providing an individual with data relating to either assessment of progression or regression of a mild cognitive impairment associated with Alzheimer’s disease or progression or regression of Alzheimer’s disease comprising:a server connected to a communication network comprising a microprocessor executing computer executable instructions stored within a memory accessible to the server which cause the server to execute a process comprising:retrieving a subset of multi-omic, including metabolomic and proteomic, data from another memory accessible to the server;processing the retrieved subset of multi-omics data with a trained computational model stored within a further memory accessible to the server; andpush data established by the trained computational model to an electronic device associated with the individual where the data is established in dependence upon the subset of multi-omics data processed with the trained computational model;the other memory accessible to the server storing biomarker data relating to the individual for a plurality of proteomic and metabolomic biomarkers established via high-throughput spectroscopic assay of one or more samples obtained from the individual;the further memory accessible to the server storing a trained computational model trained with proteomic and metabolic datasets obtained from a plurality of samples from a plurality of subjects where a first subset of the subjects have Alzheimer’s disease and a second subset have the mild cognitive impairment associated with Alzheimer’s disease; andthe electronic device comprising another microprocessor executing computer executable instructions stored within a device memory accessible to the electronic device which cause the electronic device to execute a process comprising:render upon a display of the electronic device a graphical user interface comprising a portal relating to an assessment of the individual with respect to the progression orregression of the mild cognitive impairment associated with Alzheimer’s disease or the progression or regression of Alzheimer’s disease based upon the subset of multi - omics data of the individual processed with the trained computational model; whereinthe portal renders a subject profile comprising:one or more icons where each icon of the one or more icons assigns an initial dataset from the individual to at least one of a number of endotype classes relating to progression of Alzheimer’s disease;one or more other icons where each other icon of the one or more other icons is associated with the portal providing to the individual treatment data for a treatment pathway to treat an endotype class of the number of endotype classes associated with an icon of the one or more icons; andoptionally a biomarker display comprising data relating to one or more biomarkers of the plurality of biomarkers which are out of a defined range for the number of endotype classes established by the trained computation model for the individual.

24. The system according to claim 23, whereinat least one of:the trained computational model establishes an endotype class of the number of endotype classes in dependence upon at least a defined percentage of the plurality of biomarkers are out of a normal range; andthe subset of multi-omics data comprises a first subset established in dependence upon processing one or more samples acquired from the individual and processed by an assay and a second subset established in dependence upon data acquired from one or more sensors associated with the individual.

25. The system according to claim 23, whereinthe trained computational model establishes an endotype class of the number of endotype classes in dependence upon scores of at least a defined percentage of the plurality of biomarkers meeting a predetermined condition; andeach score is established by one of:assigning a specific value of a set of specific values;normalized to a defined range;normalized in dependence upon a magnitude of another biomarker of the plurality of biomarkers;normalized in dependence upon an average or a weighted average of a set of biomarkers of the plurality of biomarkers;in dependence upon a deviation from a band defining a normal or nominal range; linearly or non-linearly defined based upon a magnitude of the biomarker.

26. The system according to claim 23, whereinthe trained computational model establishes an endotype class of the number of endotype classes in dependence upon scores of at least a defined percentage of the plurality of biomarkers meeting a predetermined condition; andeach score is established in dependence upon a nominal value or nominal band where the nominal value of nominal band is one of:constant for all individuals;constant for all individuals within a demographic to which the individual belongs; and established in dependence upon sensor data acquired from one or more sensors associated with the individual.

27. The system according to claim 23, whereinthe trained computational model establishes an endotype class of the number of endotype classes in dependence upon at least a defined percentage of the plurality of biomarkers are out of a normal range;a first endotype of the number of endotypes is characterized by sphingolipid dysregulation; a second endotype of the number of endotypes is characterized by amino acid and metabolism dysregulation; anda third endotype of the number of endotypes is characterized by inflammation and oxidative stress.

28. The system according to claim 23, whereinthe portal further comprises a shared sign-up link allowing the individual to enable at least another party to view the subject profile.

29. The system according to claim 23, whereinthe trained computational model establishes an endotype class of the number of endotype classes in dependence upon at least twenty percent (20%) of the plurality of biomarkers are out of a normal range;a first endotype of the number of endotypes is characterized by sphingolipid dysregulation; a second endotype of the number of endotypes is characterized by amino acid and metabolism dysregulation; anda third endotype of the number of endotypes is characterized by inflammation and oxidative stress.

30. The system of any one of claims 23 to 29, wherein the trained computational model has been trained by executing multimodal batch processing operations by concurrently processing batches of the proteomic and metabolomic data, and wherein the multimodal batch processing reduces computation latency relative to sequential processing.

31. A method of providing an individual with data relating to either assessment of progression or regression of a mild cognitive impairment associated with Alzheimer’s disease or progression or regression of Alzheimer’s disease comprising:providing a server connected to a communication network comprising a microprocessor executing computer executable instructions stored within a memory accessible to the server which cause the server to execute a process comprising:retrieving a subset of multi-omics data from an other memory accessible to the server; processing the retrieved subset of multi-omics data with a trained computational model stored within a further memory accessible to the server; andpush data established by the trained computational model to an electronic device associated with the individual where the data is established in dependence upon the subset of multi-omics data processed with the trained computational model;providing the other memory accessible to the server storing biomarker data relating to the individual for a plurality of proteomic and metabolomic biomarkers established via high- throughput spectroscopic assay of one or more samples obtained from the individual; providing the further memory accessible to the server storing a trained computational model trained with proteomic and metabolic datasets obtained from a plurality of samples from a plurality of subjects where a first subset of the subjects have Alzheimer’s disease and a second subset have the mild cognitive impairment associated with Alzheimer’s disease; andproviding the electronic device comprising another microprocessor executing computer executable instructions stored within a device memory accessible to the electronic device which cause the electronic device to execute a process comprising:render upon a display of the electronic device a graphical user interface comprising a portal relating to an assessment of the individual with respect to the progression or regression of the mild cognitive impairment associated with Alzheimer’s disease or the progression or regression of Alzheimer’s disease based upon the subset of multi - omics data of the individual processed with the trained computational model; whereinthe portal renders a subject profile comprising:one or more icons where each icon of the one or more icons assigns an initial dataset from the individual to at least one of a number of endotype classes relating to progression of Alzheimer’s disease;one or more other icons where each other icon of the one or more other icons is associated with the portal providing to the individual treatment data for a treatment pathway to treat an endotype class of the number of endotype classes associated with an icon of the one or more icons; andoptionally a biomarker display comprising data relating to one or more biomarkers of the plurality of biomarkers which are out of a defined range for the number of endotype classes established by the trained computation model for the individual.

32. A method of providing an individual with data relating to either assessment of progression or regression of a mild cognitive impairment associated with Alzheimer’s disease or progression or regression of Alzheimer’s disease comprising:providing a non-transitory memory accessible to a server which when executed by a processor of the server cause the server to execute a process comprising:retrieving a subset of multi-omics data from another memory accessible to the server; processing the retrieved subset of multi-omics data with a trained computational model stored within a further memory accessible to the server; andpush data established by the trained computational model to an electronic device associated with the individual where the data is established in dependence upon the subset of multi-omics data processed with the trained computational model; whereinthe other memory accessible to the server stores biomarker data relating to the individual for a plurality of proteomic and metabolomic biomarkers established via high-throughput spectroscopic assay of one or more samples obtained from the individual;the further memory accessible to the server stores a trained computational model trained with proteomic and metabolic datasets obtained from a plurality of samples from a plurality of subjects where a first subset of the subjects have Alzheimer’s disease and a second subset have the mild cognitive impairment associated with Alzheimer’s disease.

33. The method according to claim 32, whereinthe electronic device comprises another microprocessor executing computer executable instructions stored within a device memory accessible to the electronic device which cause the electronic device to execute a process comprising:render upon a display of the electronic device a graphical user interface comprising a portal relating to an assessment of the individual with respect to the progression or regression of the mild cognitive impairment associated with Alzheimer’s disease or the progression or regression of Alzheimer’s disease based upon the subset of multi - omics data of the individual processed with the trained computational model; whereinthe portal renders a subject profile comprising:one or more icons where each icon of the one or more icons assigns an initial dataset from the individual to at least one of a number of endotype classes relating to progression of Alzheimer’s disease;one or more other icons where each other icon of the one or more other icons is associated with the portal providing to the individual treatment data for a treatment pathway to treat an endotype class of the number of endotype classes associated with an icon of the one or more icons; andoptionally a biomarker display comprising data relating to one or more biomarkers of the plurality of biomarkers which are out of a defined range for the number of endotype classes established by the trained computation model for the individual.

34. The method according to claim 32, whereinthe trained computational model establishes an endotype class of the number of endotype classes in dependence upon scores of at least a defined percentage of the plurality of biomarkers meeting a predetermined condition; andeach score is established by one of:assigning a specific value of a set of specific values;normalized to a defined range;normalized in dependence upon a magnitude of another biomarker of the plurality of biomarkers;normalized in dependence upon an average or a weighted average of a set of biomarkers of the plurality of biomarkers;in dependence upon a deviation from a band defining a normal or nominal range; linearly or non-linearly defined based upon a magnitude of the biomarker.

35. The method according to claim 32, whereinthe trained computational model establishes an endotype class of the number of endotype classes in dependence upon scores of at least a defined percentage of the plurality of biomarkers meeting a predetermined condition; andeach score is established in dependence upon a nominal value or nominal band where the nominal value of nominal band is one of:constant for all individuals;constant for all individuals within a demographic to which the individual belongs; and established in dependence upon sensor data acquired from one or more sensors associated with the individual.

36. The method according to claim 32, whereinat least one of:the trained computational model establishes an endotype class of the number of endotype classes in dependence upon at least a defined percentage of the plurality of biomarkers are out of a normal range; andthe subset of multi-omics data comprises a first subset established in dependence upon processing one or more samples acquired from the individual and processed by an assay and a second subset established in dependence upon data acquired from one or more sensors associated with the individual.

37. The method of any one of claims 32 to 36, wherein the trained computational model has been trained by executing multimodal batch processing operations by concurrently processing batches of proteomic and metabolomic data, and wherein the multimodal batch processing reduces computation latency relative to sequential processing.

39. A method for producing a trained computational model for assigning a patient sample into at least one of three Alzheimer endotypes, the method comprising executing multimodal batch processing operations by concurrently processing batches of the proteomic and metabolomic data sets obtained from a plurality of samples from Alzheimer’s disease patients and normal individuals, and wherein the multimodal batch processing reduces computation latency relative to sequential processing.

Images