Systems and methods for designing and conducting clinical trials and biomarker validation studies

Abstract

In some embodiments, the subject matter of the present disclosure relates to systems and methods for designing and conducting clinical trials and biomarker validation studies. An exemplary method includes: obtaining access to a computer-implemented model; providing a trained computer-implemented model with a set of values ​​for a set of features, each set of values ​​corresponding to a candidate individual from a set of candidate individuals; receiving a disease state prediction from the trained computer-implemented model for each set of values; identifying a group of participants from the set of candidate individuals based on the disease state prediction; and facilitating at least one clinical trial or biomarker validation study in which the group of participants participate.

Description

TECHNICAL FIELD

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of U.S. Provisional Patent Application No. 63 / 584,033 filed on September 20, 2023, the entire contents of which are incorporated herein by reference in their entirety.

BACKGROUND ART

[0002] Clinical trials are essential for improving medical outcomes and advancing medical knowledge. However, a significant challenge in conducting clinical trials is the recruitment of appropriate participants. Recruiting suitable participants is extremely critical to ensuring the success and validity of clinical trials. Despite this importance, current recruitment methods often face several challenges, such as high costs, inefficiency, and difficulties in finding and engaging eligible participants.

[0003] Similarly, biomarker validation studies are extremely important in the development of personalized medicine, therapeutic interventions, and diagnostic tools. However, similar to clinical trials, a major challenge in conducting biomarker validation studies is the recruitment of appropriate candidates. Biomarker validation may require participants meeting specific criteria, such as specific genetic profiles, disease stages, or pre-existing medical conditions.

SUMMARY OF THE INVENTION

PROBLEMS TO BE SOLVED BY THE INVENTION

[0004] There is a need for improved systems and methods for recruiting candidate subjects for clinical trials, biomarker validation studies, or other longitudinal or cross-sectional studies.

MEANS FOR SOLVING THE PROBLEMS

[0005] In various embodiments, this disclosure relates to systems and methods for designing and conducting clinical trials, biomarker validation studies, and other research. A predictive model (e.g., a machine learning model) is trained to take patient data as input and provide predictions of the patient's health status or condition as output. The health status or condition may be, for example, a prediction along a health-to-disease axis or a disease risk axis, and / or may be expressed as a probability distribution. Similar predictions may be possible for other patients, and these predictions can be used to identify participants for a clinical trial (or other research). For example, a clinical trial may have a target stage or target probability distribution of the disease being studied, and the model predictions can be used to select patients who meet the target stage or target probability distribution. In some cases, for example, patients may be selected so that the sum of their predicted probability distributions matches or approximates the target probability distribution of the clinical trial. Other patient characteristics, such as age, sex, or place of residence, may also be considered when selecting patients. Selected patients can be enrolled as participants in the clinical trial. Advantageously, in certain embodiments, the systems and methods described herein enable more accurate and efficient identification and selection of patients who meet the requirements of a clinical trial or other research.

[0006] In one embodiment, the subject matter of the present disclosure relates to a method for conducting a clinical trial or biomarker validation study. The method includes: obtaining access to a computer-implemented model trained with training data comprising multiple records relating to multiple individuals, wherein each record comprises (i) values ​​of multiple features relating to each individual obtained from the multiple individuals, and (ii) a label indicating the pathology of each individual; providing the trained computer-implemented model with multiple sets of values ​​for the multiple features, wherein each set of values ​​corresponds to a candidate individual from a set of candidate individuals; receiving a pathology prediction from the trained computer-implemented model for each set of values; identifying a participant group from the set of candidate individuals from the pathology prediction; and facilitating at least one clinical trial or biomarker validation study in which the participant group participates.

[0007] In specific embodiments, the values ​​of multiple features describe at least one of the following: demographics, medical interviews, physical examinations, diagnoses, medications, medical procedures, vital sign measurements, vaccinations, test results, serum, urine samples, biological samples, gene expression levels, medical images, medical records, radiographic reports, genetic tests, biomarkers, pathology reports, health information, social determinants of health, financial data, consumer data, or any combination thereof. The model may include at least one of the following: regression models, classifiers, linear models, nonlinear models, random forests, kernel methods, Bayesian models, decision trees, or neural networks. The disease state prediction may include probability distributions. The disease state prediction may be based on or associated with a primary stratification axis, which may be a health-to-disease axis or a disease risk axis, or include such axes.

[0008] In some implementations, participant group identification involves stratifying each candidate individual from a group of candidates along one or more primary stratification axes. Participant group identification may further involve stratifying each candidate individual from a group of candidates along one or more secondary stratification axes. One or more secondary stratification axes may represent one or more characteristics of the group of individuals, which may include at least one of physiological comorbidities, demographics, or socioeconomic variables. At least one stratification axis from one or more primary stratification axes or one or more secondary stratification axes may represent at least one of continuous or ordinal dimensions. Locations along at least one stratification axis from one or more primary stratification axes or one or more secondary stratification axes may be represented by points, point estimates, or probability distributions.

[0009] In various cases, clinical trials or biomarker validation studies may include studying at least one of the following: the progression of health status, the validity of biomarkers, the validity of biomarker panels, the validity of targeted therapies, the effectiveness of therapeutic interventions, the effectiveness of digital interventions, the effectiveness of behavioral interventions, or any combination thereof. Participant groups may enhance one or more characteristics compared to a set of candidate individuals, such as reduced physiological heterogeneity, increased probability of outcome events occurring, increased tendency to respond to interventions, increased likelihood of observing specific physiological or pathological signs, or any combination thereof.

[0010] In some embodiments, participant group identification includes: determining a target distribution for a set of candidate individuals in accordance with the objectives of a clinical trial or biomarker validation study, wherein the target distribution defines a probability distribution; and selecting participant groups from the set of candidate individuals such that a collective distribution of participant groups similar to the target distribution is achieved. The target distribution may be determined based on at least one parameter to be evaluated in the clinical trial or biomarker validation study, the at least one parameter including at least one of biological characteristics, biomarkers, biomarker panels, therapeutic targets, or pharmacological interventions. The target distribution may be constant or uniform with respect to disease state (e.g., along a health-to-disease axis). The target distribution may include a risk probability distribution. The target distribution may be enhanced with respect to regions of a stratification axis related to disease state. The step of selecting participant groups may include minimizing the number of individuals within each participant group. The selection of participant groups may include: identifying multiple groups of individuals, each of which has a similar probability distribution with respect to the disease state; and determining the number of individuals from each group (and / or individuals not belonging to any group) to be included in a participant group in order to satisfy a target distribution.

[0011] In various cases, this method includes facilitating clinical trials, which include evaluating the effectiveness of treatments. Facilitating the implementation of clinical trials includes enrolling members from a participant group in a clinical trial, thereby providing members with drugs, digital interventions, or behavioral interventions. This method also includes facilitating the implementation of biomarker validation studies, which may include the validation of biomarkers or biomarker panels.

[0012] In various implementations, identifying participant groups involves achieving a desired distribution of at least one covariate, the at least one covariate including demographics, comorbidities, drug use, medical history, socioeconomic characteristics, biological characteristics, or any combination thereof. This method may include the step of training a computer implementation model using training data.

[0013] In another aspect, the subject matter of the present disclosure relates to a system for conducting a clinical trial or biomarker validation study. The system includes one or more computer processing units programmed to perform operations including: accessing or obtaining access to a computer-implemented model trained with training data containing multiple records relating to multiple individuals, wherein each record contains (i) multiple feature values ​​relating to each individual from multiple individuals, and (ii) a label indicating the pathology of each individual; providing the trained computer-implemented model with each set of multiple values ​​corresponding to a candidate individual from a set of candidate individuals; receiving a pathology prediction from the trained computer-implemented model for each set of values; identifying a participant group from the set of candidate individuals from the pathology prediction; and facilitating at least one clinical trial or biomarker validation study in which the participant group participates.

[0014] In certain embodiments, disease state prediction may be based on or related to a primary stratification axis, including a health-to-disease axis or a disease risk axis. Identification of participant groups may include the step of stratifying each candidate individual from a group of candidates according to one or more primary stratification axes and / or one or more secondary stratification axes.

[0015] Identifying participant groups may include: determining a target distribution for a set of candidate individuals in accordance with the objectives of a clinical trial or biomarker validation study, wherein the target distribution defines a probability distribution; and selecting participant groups from the set of candidate individuals to achieve a collective distribution of participant groups that is similar to or approximates the target distribution. The target distribution may include a risk probability distribution. Selecting participant groups may include: identifying multiple groups of individuals, each of which has a similar probability distribution with respect to the disease state; and determining the number of individuals from each group (and / or individuals not belonging to any group) to be included in participant groups to satisfy the target distribution. These operations may include training a computer-implemented model using training data.

[0016] These and other objectives, as well as the advantages and features of the embodiments of the invention disclosed herein, will become further apparent by reference to the following description, drawings, and claims. Furthermore, it should be understood that the features of the various embodiments described herein are not mutually exclusive and may exist in various combinations and permutations.

[0017] The above will become clearer from the following more specific descriptions of embodiments, as shown in the attached drawings. Note that in these drawings, similar reference symbols indicate the same elements throughout the various figures. The drawings are not necessarily to a fixed scale, and the emphasis is on describing the embodiments. [Brief explanation of the drawing]

[0018] [Figure 1A-1B] This is a plot of the number of individuals against a given disease state, along an axis from health to disease, according to a specific embodiment. [Figure 2] This is a schematic diagram of a method for identifying a cohort of individuals for a clinical trial or other study, according to a specific embodiment. [Figure 3]FIG. 2 is a schematic diagram of a method for training a machine learning model in accordance with certain embodiments. [Figure 4] FIG. 3 is a schematic diagram of a method of using a predictive model to identify a cohort of individuals eligible for participation in a study in accordance with certain embodiments. [Figure 5] FIG. 4 is a block diagram of an exemplary computer system in accordance with certain embodiments. DETAILED DESCRIPTION OF EMBODIMENTS FOR CARRYING OUT THE INVENTION

[0019] A description of example embodiments is provided below. It is contemplated that the claimed apparatus, systems, methods, and processes of the invention include variations and modifications developed using information from the embodiments described herein. Alterations and / or modifications of the apparatus, systems, methods, and processes described herein may be implemented by those of ordinary skill in the relevant technical field.

[0020] It should be understood that the order of steps or the order for performing specific operations is not critical so long as the invention remains operable. Furthermore, two or more steps or operations may be performed concurrently.

[0021] In various embodiments, a "organism" (also referred to herein as an "individual (subject)", "test subject", or "patient") may refer to an animal, plant, fungus, or any living being. The organism may be, for example, a mammal such as a human or a mouse.

[0022] In various embodiments, "disease" can refer to any disorder of function or structure in an organism. The disease can be any type of disease including, for example, cancer (e.g., colorectal cancer, lung cancer, skin cancer, etc.), genetic diseases (e.g., spinal muscular atrophy (SMA), Huntington's disease (HD), or Duchenne muscular dystrophy (DMD)), liver diseases (e.g., metabolic dysfunction-associated steatotic liver disease (MASLD), metabolic dysfunction-associated steatohepatitis (MASH), or cirrhosis), kidney diseases (e.g., chronic kidney disease (CKD) or polycystic kidney disease), diabetes (e.g., type 2 diabetes), neurodegenerative diseases (e.g., Alzheimer's disease (AD), amyotrophic lateral sclerosis (ALS), or Parkinson's disease (PD)), autoimmune diseases (e.g., multiple sclerosis (MS), rheumatoid arthritis (RA), or inflammatory bowel disease (IBD)), lung diseases (e.g., chronic obstructive pulmonary disease (COPD) or pulmonary fibrosis (e.g., IPF)), post-COVID conditions, colorectal polyps, age-related macular degeneration (AMD), or any other type of disease. In some instances, the disease may be, include, and / or present with inflammation, cellular response to stress, or other types of dysfunction.

[0023] In various embodiments, "disease of interest" can refer to a disease that is characterized or analyzed using the predictive modeling techniques described herein. For example, a machine learning model can be trained to make predictions related to the disease of interest.

[0024] In various embodiments, a "reference domain" (also referred to herein as "feature space") can be or include a collection of features or variables (e.g., biological data, latent variables, or phenotype-specific variables) that can be used as training data for training a machine learning model and / or as input to a machine learning model. The reference domain can include, for example, gene expression levels and / or other biological features associated with the disease of interest.

[0025] In various embodiments, “biomarker” can refer to a biological condition or a measurable indicator of a condition. Biomarkers can indicate the presence or progression of a disease or the effectiveness of a treatment. Biomarkers can be biomolecules and may be present in tissues, blood, or bodily fluids. Biomarkers can be used for diagnostic purposes, for example, to confirm or detect the presence of a condition or disease in question, or to identify individuals having a subtype of disease. Repeated measurements of biomarkers can be used to monitor individuals, for example, to assess the state of a medical condition or disease, or to find evidence of the effects of environmental factors or medical products (or exposure to them).

[0026] In certain embodiments, one or more predictive models (e.g., machine learning models or computer-implemented models) are used to identify individuals eligible to participate in a clinical trial, biomarker validation study, cross-sectional study, or longitudinal study (for simplicity, referred to herein as a "trial" or "study," respectively). One or more predictive models may be configured to take an individual's health data as input and provide an output of that individual's predicted disease state, health status, and / or disease risk. Such predictions can be made for a set of individuals. Based on the predictions, a cohort of individuals that meet the target disease state, target health status, and / or target disease risk can be selected for the clinical trial or other study.

[0027] Figure 1A is a plot 100 of the number of individuals N against the pathogenesis of a given disease along the health-to-disease axis, according to a specific embodiment. Plot 100 includes population distributions for two cohorts. The first population distribution 110 concerns a cohort simply selected from the general population, showing that most individuals are healthy. The second population distribution 112 concerns a cohort corresponding to a target stage 114 in a clinical trial. For example, the cohort of the second population distribution 112 may be focused on for prescribing and evaluating treatment in a clinical trial. In various embodiments, the cohort of the second population distribution 112 can be identified using the systems and methods described herein. The target stage 114 may correspond to a range or value of disease severity or risk. For example, the target stage may correspond to a numerical range or value of eGFR in chronic kidney disease, visual acuity in age-related macular degeneration, time to diagnosis in Parkinson's disease, or diabetes risk (e.g., risk of developing diabetes).

[0028] Figure 1B shows a similar plot 120 of the number of individuals N for a given disease with respect to the disease state along the health-to-disease axis, according to a specific embodiment. Plot 120 includes a first population distribution 110 corresponding to a simply selected cohort. Plot 120 also includes a third population distribution 122 showing a uniform or homogeneous distribution across the health-to-disease axis. In various embodiments, the third population distribution 122 is desirable for biomarker validation studies where the reliability and accuracy of biomarkers can be evaluated.

[0029] In certain embodiments, cohorts from the second population distribution 112 and the third population distribution 122 can be identified using the systems and methods described herein. For example, a machine learning model (or other predictive model) can be configured to take an individual's health information as input and determine an individual's disease state or health condition as output. To form a cohort of the second population distribution 112, the machine learning model can be used to identify individuals with a disease state corresponding to the target stage 114 targeted in the clinical trial. Such individuals can be included in the cohort, while other individuals with disease states outside the target stage 114 can be excluded from the cohort.

[0030] Similarly, to create a cohort of the third population distribution 122, a machine learning model can be used to identify and select individuals with various disease states ranging from healthy to unhealthy across the health-to-disease axis. In some embodiments, a uniform representation of the third population distribution 122 can be obtained by selecting combinations of healthy individuals (e.g., low disease risk), individuals in moderate health (e.g., moderate disease risk), unhealthy individuals (e.g., high disease risk), and / or other individuals in intermediate stages of illness. The combinations of disease states of the selected individuals can achieve or approximate a uniform representation of the third population distribution 122.

[0031] In various embodiments, the selection of individuals for a cohort can be carried out by considering each individual's risk probability distribution. For example, each individual may have a risk probability distribution along a health-to-disease axis, and a cohort can be selected such that the sum of the risk probability distributions for individuals within the cohort equals a desired population distribution (or an approximation thereof), such as a third population distribution.122 In some embodiments, the risk probability distribution provides the probability associated with an individual having or developing a disease. In addition to, or instead of, the risk probability distribution can be correlated with, or derived from, an individual's predicted health status along a health-to-disease axis. For example, an individual predicted to be healthy may have a risk probability distribution indicating a low risk of having or developing a disease. Similarly, an individual predicted to be unhealthy may have a risk probability distribution indicating a high risk of having or developing a disease. Diseases may have multiple risk levels, such as low, medium-low, medium-high, or high, and the risk probability distribution can assign probability values ​​for each risk level. Risk probabilities and risk probability distributions can be calculated using the predictive models described herein.

[0032] Figure 2 is a schematic diagram of a method 200 for identifying a desired cohort according to a specific embodiment. Using a predictive model, risk probability distributions 210, 212, and 214 are calculated for individual patients. In the illustrated embodiment, patients are broadly classified into three groups: a low-risk group represented by risk probability distribution 210 (having a downward-shifted risk distribution), a medium-risk group represented by risk probability distribution 212 (having a risk distribution with a peak in the central region), and a high-risk group represented by risk probability distribution 214 (having an upward-shifted risk distribution). A certain number of patients can be selected or sampled from each group to constitute the final cohort. In the illustrated embodiment, 10 patients are sampled from the low-risk group, 8 from the medium-risk group, and 12 from the high-risk group. In addition to or instead of this, if desired, one or more patients with unique risk probability distributions (e.g., different from risk probability distributions 210, 212, and 214) can also be selected. The cohort obtained in the illustrated embodiment consists of low-risk, medium-risk, and high-risk subjects. The cohort histogram 216 shows the estimated number of patients in each of the following risk categories: low, moderate-low, moderate-high, and high. The threshold line 218 corresponds to the target distribution of the study containing the cohort, for example, as specified by the study sponsor.

[0033] In some embodiments, "surplus" subjects can be recruited to ensure that the total density or number of patients for all risk categories exceeds the threshold 218. For example, the minimum number of patients that satisfies or approximates the target distribution and has a threshold number of patients within each risk category can be selected. Advantageously, this approach allows for the selection of a recruitment cohort that satisfies the target parameters from the risk probability distribution, even when the model cannot precisely place individual subjects on the health-to-disease axis. While the target distribution (represented by the threshold line 218) in this example is constant or uniform, it is understood that the target distribution can also be heterogeneous. For example, the target distribution may be similar to the risk probability distributions 210, 212, or 214.

[0034] In some embodiments involving biomarker validation, it may be desirable to obtain a uniform distribution of individuals along a health-to-disease axis, a disease risk axis, or other dimensions (e.g., those represented by a third population distribution 122 or threshold line 218). Advantageously, Method 200 can achieve the desired distribution by utilizing predictive modeling and sampling based on probability distributions. In specific embodiments, one or more models (referred to here as “models” for simplicity) generate probability distributions representing each individual’s health status or risk probability, and the models can aggregate individuals to create a broader target population distribution. This approach can provide an expected number of subjects or predicted size for the recruited cohort. The number of subjects can be used to evaluate the quality of Method 200 or other methods used to identify the cohort. Generally, more robust methods can meet trial requirements with a smaller number of subjects.

[0035] In some embodiments, one or more stratification axes can be selected for the group of individuals being considered for study (referred to as “candidate individuals”), and the model can stratify each individual within the group according to one or more stratification axes. The stratification axes can define, relate to, and / or provide measures for one or more inclusion criteria for the study. Such criteria may include one or more primary stratification axes and, optionally, one or more secondary stratification axes. One or more primary stratification axes may relate to or provide measures for health or disease, such as disease state or stage (e.g., along the health-to-disease axis), health status, and / or disease risk (e.g., risk of having a disease or risk of developing it along the disease risk axis). One or more secondary stratification axes may include or relate to other characteristics, such as sex, age, ethnicity, demographics, treatment history, drug use, physiological comorbidities, socioeconomic variables, or any combination thereof. For example, in addition to recruiting participants evenly along a health-to-disease axis (e.g., primary stratification axis), the trial designer (e.g., sponsor) may also want an equal number of male and female participants (e.g., secondary stratification axis). In the specific example of chronic kidney disease, it may be desirable to select participants with a specific distribution of eGFR values ​​(e.g., primary stratification axis of eGFR) and a specific distribution of age, sex, and / or distance to the clinical site (three secondary stratification axes).

[0036] In various embodiments, the model and / or related techniques can be used to identify patients to be recruited for research based on their location or probability distribution along selection axes. Using the model, a subgroup of candidate individuals can be selected and / or identified whose collective distribution of location or probability along one or more stratification axes is predicted to resemble or approximate the target distribution of the study (e.g., a second or third population distribution 112 or 122). Using such predictions, individuals selected for research can be recruited, for example, through the automated generation of invitation letters, emails, or notifications to healthcare professionals via an electronic medical record system. Alternatively, or in addition, selected individuals may also be recruited through primary care physicians, specialist healthcare professionals, or other settings. In some embodiments, information related to the stratification axes can be obtained from a database of records about candidate individuals.

[0037] In addition to or instead of this, treatment decisions for a patient can be made using risk probabilities or risk probability distributions. For example, patient data can be provided to a predictive model described herein, which can then determine and / or output a probability distribution (e.g., a risk probability distribution) representing the patient's health status. This probability distribution can be used by the patient, physician, and / or other healthcare professionals to make decisions regarding the patient's medical care and / or treatment. In some embodiments, the predictive model can use the probability distribution to determine and / or output one or more treatment recommendations for the patient. In some embodiments, the probability distribution can be used to derive a measure (such as disease risk or probability (%)) that is familiar and interpretable to the patient or physician.

[0038] Figure 3 is a schematic diagram of a method 300 for training a machine learning model 310 according to a particular embodiment. In some embodiments, model 310 may be or include a regression model, classifier, linear model, nonlinear model, random forest, kernel method, Bayesian model, decision tree (e.g., XGBoost), neural network, generative model, or other type of machine learning model. In a particular embodiment, model 310 may be, for example, a multilayer fully connected neural network.

[0039] In various embodiments, Model 310 can be trained to accept data relating to an individual as input and to provide predicted conditions, health status, and / or disease risks for that individual as output. The input data used to train Model 310 may include, for example, patient data 312 (e.g., medications, diagnoses, tests, vital signs, etc.) and / or survey data 314 (if available). The model inputs can form the reference domain or feature space of Model 310. In some embodiments, patient data 312 may include real-world data that exists in medical records or can be created using tests, imaging, other tests, consumer data, social determinants of health, genetic data, or other relevant information sources. Such data can be measured directly or estimated (e.g., using regression models). Survey data 314 may represent additional prospectively collected data for each subject that can be obtained from questions and corresponding answers and / or may not be readily available from medical records or other data sources. Survey data 314 and / or patient data 312 may include, for example, waist-to-hip ratio, waist circumference, hip circumference, body fat percentage, blood pressure, employment status, diet, exercise, etc. In some cases, model input data may include demographic information (e.g., sex, ethnicity, income, etc.).

[0040] In certain embodiments, the training data and / or model output may include a sequential representation of health status 316, which may be, or include, predicted values ​​or scores along a continuous dimension of health status (e.g., a health-to-disease axis). In certain embodiments, the sequential representation of health status 316 may form the basis of a current disease staging paradigm (typically abstracted into higher-level categories such as "Stage II" or "Stage IV"). The sequential representation of health status 316 may serve as labels or ground truth data in the training data. The sequential representation of health status 316 may be, or include, image-based fat percentage in the case of non-alcoholic fatty liver disease (NAFLD), estimated glomerular filtration rate (eGFR) in the case of chronic kidney disease (CKD), and / or Unified Parkinson's Disease Rating Scale (UPDRS) in the case of Parkinson's disease (PD).

[0041] Other features can be used as model inputs and / or model outputs (e.g., in the training data). For example, model inputs and / or model outputs may include any dimensions or variables measured or predicted in accordance with the trend of the transition, the time to the transition, the biological stage, the physiological stage, and / or the transition from health to disease. For example, in the case of PD, model inputs may include brain imaging, expert functional assessments, diagnostic records, social determinants of health, or any combination thereof. For example, in the case of diabetes, model inputs may include data on dietary records, lifestyle factors, medical history, blood test values, or a combination thereof. Model outputs may include estimated time to diagnosis (e.g., in the case of PD) and / or the risk of developing the disease (e.g., diabetes). The risk of developing the disease can be stratified into high, medium, and low-risk cohorts.

[0042] Using sufficient ground truth data, models can be trained to analyze individuals who have or do not have a disease or condition. This enables interpretability and / or reusability of information learned across and / or over time within healthcare systems, and allows models to make predictions about data-naive individuals (e.g., those not yet analyzed by the system or not used as training data sources).

[0043] In some cases, Model 310 can be trained to selectively reinforce a desired population. This may include steps such as separating cohorts of individuals across relevant dimensions (e.g., health status), precisely defining the cohorts at or near specific locations along those dimensions, and / or identifying patterns of features that recognize the cohorts. Cohorts can be selected to participate in clinical trials or other studies as described herein.

[0044] In some cases, the data used to generate training data or model inputs (e.g., test results, imaging results, and / or other patient data) may lack consistency in formatting, language, and / or terminology. Mapping such data to a unified or consistent format, language, and / or terminology can be beneficial for model training and use.

[0045] In various embodiments, this model can be used to determine when different therapeutic targets or interventions are relevant and / or effective. For example, some interventions that inhibit lipid accumulation can be delivered in the earliest stages of NAFLD, while interventions that inhibit factors that mobilize immune cells or activate fibrotic cells can be applied in later stages of progression. One or more predictive biomarkers can be used (e.g., by the model) to identify individuals in a specific stage of the disease and / or individuals who are more likely than others to experience favorable or unfavorable effects from a therapeutic target or intervention.

[0046] In some embodiments, continuous time scales can be generated and used using the systems and methods described herein. In some embodiments, labels for novel or different subpopulations can be provided by training a model with labels from subpopulations. Training data can be generated by extracting PDFF (e.g., fat content) from individuals with MRI data and / or obtaining eGFR from individuals with standard clinical laboratory values. Using these generated or extracted labels or values, a model can be trained to learn how to predict these features in other individuals based on traditional data that is readily available or passively collected (e.g., derived from electronic health records and / or surveys).

[0047] In some embodiments, two approaches can be used to identify and validate biomarkers. The first approach uses biological studies to generate snRNA-seq data from tissues collected across a spectrum from health to disease (e.g., multiple molecular stages), and then identifies proteins that are likely to be secreted and detected in the blood. The second approach involves collecting blood from individuals recruited at various stages in a continuous spectrum from health to disease. Studies can then be conducted on the blood to identify or determine changes associated with the progression of the disease. The studies may be biased, unbiased, proteomics, metabolomics, lipidomics, or any combination thereof.

[0048] Figure 4 is a schematic diagram of a method 400 for identifying a cohort of individuals eligible to participate in a clinical trial or study using a predictive model (e.g., Model 310) according to a specific embodiment. Health data, demographic data, and / or survey data (e.g., patient data 312 and survey data 314) relating to an individual 412 are provided as input to the predictive model 310. The data may be entered via a user interface by the individual 412, a physician, a healthcare professional, a clinical trial designer, or another, or the data may be entered by automated data pulling via an electronic health record (EHR) system. As output, Model 310 can provide predictions of an individual's health status (e.g., a sequential representation of health 316), pathology, and / or disease risk. For example, Model 310 can predict for an individual 412 the values ​​or states of molecular markers (e.g., gene expression levels) across a spectrum from health to disease. Similar predictions can be made for other individuals. In some embodiments, the health-to-disease spectrum can capture the time course of impairment without the need to longitudinally track the individual.

[0049] In the illustrated embodiment, individual 412's condition 414 falls within the range of a target condition 416, which is situated between a healthy state and a diseased state. The target condition 416 may correspond to the target population intended for the clinical trial. For example, the target condition 416 can be selected to capture individuals who are most likely to respond to the treatment relevant to the clinical trial. In many chronic diseases, such as diabetes mellitus, IBD, or age-related macular degeneration (AMD), different treatments may be used for patients depending on the severity of the disease (e.g., early, moderate, or severe). Once the target condition 416 is identified, a cohort of individuals with a health condition (e.g., condition 414) that falls within the range of the target condition 416 is identified using model predictions. Alternatively, or in addition to this, a cohort may also be identified using a probability distribution (e.g., risk probability distributions 210, 212, and / or 214) that satisfies the target condition 416 or approximates the target distribution, as described herein. Such individuals can then be recruited for the clinical trial, as described herein. Once individuals have been recruited, a clinical trial can be conducted by providing some or all of the recruited individuals with an intervention, treatment, or procedure 418. For example, the first group of individuals in the test group may be given the procedure 418, and the second group of individuals in the control or placebo group may be given a placebo. The results of the trial can be used to determine whether the procedure shows positive results for the individuals who received it.

[0050] In certain embodiments, Model 310 can provide predictions of disease risk for individuals 412 and other candidate individuals being considered for inclusion in clinical trials. Disease risk may correspond to a transition from a healthy state to a diseased state. This transition may include, for example, an early biodynamic shift, a mid-term biodynamic shift, or a late biodynamic shift. Such biodynamic shifts may be cascades with different characteristics in the early, mid, and late stages of the disease. Using a cohort of individuals identified for study inclusion, a relationship 420 between the levels of molecular markers and the biodynamic shift from a healthy state to disease can be determined. From the relationship 420, trends for all sampled individuals can be identified. In the illustrated embodiment, the target state 416 corresponds to a mid-term biodynamic shift. Since these individuals are more likely to experience disease progression than healthy individuals, if a therapeutic effect exists, it may be easier to observe that effect during the study period.

[0051] Embodiments, systems, and methods described herein, including predictive models, can be implemented via a network (e.g., including the Internet). For example, a predictive model can be accessed by a clinical trial designer via the network, where it identifies or provides a cohort of individuals via a HIPAA-compliant and secure connection. Furthermore, the cohort is provided in a double-blind format, and participants can be contacted to register for the study, but their identity is masked from the scientists conducting the study.

[0052] In various embodiments, the systems and methods described herein can be used to facilitate clinical trials, biomarker validation studies, or other research. Facilitating a clinical trial (or other research) may include a variety of activities related to the organization or conduct of the trial. Such activities may include, for example: recruiting a group of participants to be studied (for example, identified using the systems and methods described herein); enrolling or registering a group of participants in the trial or study; collecting biological samples, image data, or other data as part of the trial or study; treating one or more participants in the trial or study with interventions; applying diagnoses to one or more participants; making recommendations regarding treatment decisions for one or more participants; making recommendations regarding diagnostic or screening tests for one or more participants; or any combination thereof.

[0053] Machine learning and neural networks Machine learning is a method of training computers to learn and make decisions on their own, without being explicitly programmed to perform specific tasks. When a computer program is fed a large amount of data, it uses statistical analysis to identify patterns and relationships within the data. The goal is to enable the program to make predictions or decisions based on these patterns and relationships without being explicitly instructed how to do so.

[0054] A neural network is a type of machine learning algorithm inspired by the structure and function of the human brain. A neural network consists of layers of interconnected "neurons" (sometimes called nodes), which process and transmit information. Each neuron receives input from other neurons, processes it, and then transmits the information to the neurons in the next layer.

[0055] In a neural network, a layer refers to a layer of interconnected neurons. A neural network typically has multiple layers: the input layer receives raw data, and the output layer generates the final prediction or decision. Between the input and output layers, there are one or more hidden layers that process the data and pass it on to the next layer.

[0056] By training a neural network on a large dataset, we can adjust the connections between neurons (called "weights") to improve the network's ability to make predictions or decisions. To train a neural network, data is fed through the network, and its output is compared to the desired result. If the output is inaccurate, the weights are adjusted to reduce the error. This process is repeated multiple times, and the network continuously adjusts its weights to improve accuracy. Once the network is trained, it can be used to make predictions and decisions on new data based on the patterns and relationships learned from the training data.

[0057] In various embodiments, “machine learning” may refer to the application of specific techniques (e.g., pattern recognition and / or statistical inference techniques) by a computer system to perform a particular task. Machine learning techniques (automated or otherwise) can be used to build a data analysis model based on sample data (e.g., “training data”) and to validate the model using validation data (e.g., “test data”). Sample and validation data may be organized as a set of records (e.g., “observations” or “data samples”), each record representing a value for a specified data field (e.g., “independent variable”, “input”, “feature”, or “predictor variable”) and a corresponding value for another data field (e.g., “dependent variable”, “output”, or “target”). Machine learning techniques can be used to train a model to estimate output values ​​based on input values. Given other data similar to or related to the sample data (e.g., “inference data”), such a model can accurately estimate unknown target values ​​for the inference dataset. In this specification, such a model may be referred to as a “machine learning model”, a “predictive model”, or a “computer-implemented model”.

[0058] Features of a data sample can be measurable properties of the entity represented by or associated with that data sample (e.g., a cell, a biological sample, a person, an object, an event, an activity, etc.). For example, a feature could be a property of a cell in an organism. Another example is that a feature could be a gene expression level associated with a cell. In some cases, features of a data sample are descriptions of the entity represented by or associated with that entity (or other information about it). The value of a feature can be a measurement of a corresponding property of an entity, or an instance of information about an entity. For example, in the above example where the cell feature is a gene expression level, the value of the feature "expression of gene G" could be 215,000, which is the number of messenger RNA fragments in that cell that map to the region of gene G on the human reference genome. In some cases, the value of a feature may be missing (e.g., no value). For example, in the above example where the feature is a gene expression level, the value of that feature could be "NULL," which indicates that the gene level could not be measured using the given technique.

[0059] Features can also have data types. For example, features can be image data types, numerical data types, text data types (e.g., structured text data types or unstructured ("free") text data types), categorical data types, or other appropriate data types. In the example above, the shape feature extracted from the cell image could be of image data type. Generally, if the set of values ​​that can be assigned to a feature is finite, the data type of that feature is categorical.

[0060] As used herein, “development” of a machine learning model may refer to the construction of a machine learning model. A machine learning model can be constructed by a computer using a training dataset. Therefore, “development” of a machine learning model may include training the machine learning model using a training dataset. In some cases (generally called “supervised learning”), the training dataset used to train a machine learning model may contain known outcomes (e.g., labels or target values) for individual data samples within the training dataset. For example, when training a supervised computer vision model to detect images of cats, the target value for a data sample in the training dataset may indicate whether or not that data sample contains an image of a cat. In other cases (generally called “unsupervised learning”), the training dataset does not contain known outcomes for individual data samples within the training dataset.

[0061] After development, the machine learning model can be used to generate inferences on an “inference” dataset. For example, after development, the computer vision model may be configured to distinguish data samples containing images of cats from data samples that do not contain images of cats. As used herein, “deployment” of a machine learning model may refer to using the developed machine learning model to generate inferences on data other than the training data.

[0062] Computer implementation In some embodiments, some or all of the processes described above can be performed on a personal computer device, one or more centralized computer devices, or via cloud-based processing by one or more servers. It is also possible to perform one type of processing on one device and another type of processing on another. Some or all of the data described above can be stored on a personal computer device, in data storage hosted on one or more centralized computer devices, and / or via cloud-based storage. Some data can be stored in one location, and other data in another. In some embodiments, quantum computing can be used, and / or functional programming languages ​​can be used. Electrical memory, such as flash-based memory, can be used.

[0063] Figure 5 is a block diagram showing an example of a computer system 500 that may be used to carry out the technology described herein. A general-purpose computer, network appliance, mobile device, or other electronic system may also include at least a portion of the system 500. The system 500 includes a processor 510, memory 520, storage device 530, and input / output device 540. Each of the components 510, 520, 530, and 540 may be interconnected, for example, using a system bus 550. The processor 510 can process instructions executed within the system 500. In some implementations, the processor 510 is a single-threaded processor. In some implementations, the processor 510 is a multi-threaded processor. The processor 510 can process instructions stored in memory 520 or storage device 530.

[0064] Memory 520 stores information within the system 500. In some implementations, memory 520 is a non-temporary computer-readable medium. In some implementations, memory 520 is a volatile memory unit. In some implementations, memory 520 is a non-volatile memory unit.

[0065] The storage device 530 can provide mass storage to the system 500. In some implementations, the storage device 530 is a non-temporary computer-readable storage medium. In various different implementations, the storage device 530 may include, for example, a hard disk drive, an optical disk drive, a solid-state drive, a flash drive, or other mass storage devices. For example, the storage device may store long-term storage data (such as database data or file system data). The input / output device 540 provides input / output operations to the system 500. In some embodiments, the input / output device 540 may include one or more network interface devices (e.g., an Ethernet® card), a serial communication device (e.g., an RS-232 port), and / or wireless interface devices (e.g., an 802.11 card, a wireless modem (e.g., 3G, 4G, or 5G)). In some implementations, the input / output device may include a driver device configured to receive input data and transmit output data to other input / output devices (e.g., a keyboard, printer, and display device 560). In some embodiments, mobile computing devices, mobile communication devices, and other devices may be used.

[0066] In some implementations, at least part of the approach described above can be implemented by instructions that cause one or more processing units to perform the processes and functions described above at runtime. Such instructions may include, for example, interpretable instructions such as script instructions, executable code, or other instructions stored on non-temporary computer-readable media. The storage device 530 may be implemented, for example, distributed across a network as a server farm or a set of widely distributed servers, or it may be implemented on a single computing device.

[0067] An exemplary processing system is illustrated in Figure 5, but embodiments of the subject matter, functional operations, and processes described herein may be implemented in other types of digital electronic circuits, tangibly implemented computer software or firmware, computer hardware (including structures disclosed herein and their structural equivalents), or one or more combinations thereof. Embodiments of the subject matter described herein may be implemented as one or more computer programs, i.e., one or more computer program instruction modules encoded in a tangible non-volatile program carrier for the purpose of execution by a data processing device or for controlling its operation. Alternatively or in addition, program instructions may be encoded in artificially generated propagating signals, e.g., machine-generated electrical, optical, or electromagnetic signals, which generate information for transmission to a suitable receiving device for the purpose of execution by the data processing device. The computer storage medium may be a machine-readable storage device, a machine-readable storage board, a random or serial access memory device, or one or more combinations thereof.

[0068] The term "system" may encompass all kinds of devices, equipment, and machines for data processing, such as programmable processors, computers, or systems consisting of multiple processors or computers. A processing system may include, for example, dedicated logic circuits such as FPGAs (Field Programmable Gate Arrays) or ASICs (Application-Specific Integrated Circuits). In addition to hardware, a processing system may include code that constitutes the execution environment of the computer program, such as processor firmware, protocol stacks, database management systems, operating systems, or one or more combinations thereof.

[0069] Computer programs (sometimes also called programs, software, software applications, engines, pipelines, modules, software modules, scripts, or code) can be written in any form of programming language, including compiled languages, interpreted languages, declarative languages, or procedural languages, and can be deployed in any form, either as standalone programs or as modules, components, subroutines, or other units suitable for use in a computing environment. Computer programs can, but are not required, correspond to files in a file system. A program can be stored in part of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program, or in multiple collaborative files (e.g., a file that stores one or more modules, subprograms, or parts of code). Computer programs can be deployed to run on a single computer, on multiple computers located in a single site, or on multiple computers distributed across multiple sites and interconnected by a communication network.

[0070] The processes and logic flows described herein can be implemented by one or more programmable computers, which perform their functions by executing one or more computer programs to manipulate input data and generate outputs. Alternatively, the processes and logic flows can also be implemented by dedicated logic circuits, such as FPGAs (Field-Programmable Gate Arrays) or ASICs (Application-Specific Integrated Circuits), and the device can be implemented as such circuits.

[0071] A computer suitable for running computer programs may include, for example, a general-purpose or dedicated microprocessor, or both, or any other type of central processing unit. Generally, the central processing unit receives instructions and data from read-only memory or random-access memory, or both. A computer generally includes a central processing unit for executing or running instructions and one or more memory devices for storing instructions and data. Generally, a computer also includes one or more mass storage devices (e.g., magnetic disks, magneto-optical disks, or optical disks) for storing data, or is operablely coupled to them to receive data from, transfer data to, or both. However, a computer does not necessarily have to include such devices. Furthermore, computers can be incorporated into other devices, such as mobile phones, personal digital assistants (PDAs), portable audio / video players, game consoles, Global Positioning System (GPS) receivers, or portable storage devices (e.g., Universal Serial Bus (USB) flash drives), but these are just a few examples.

[0072] Computer-readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media, and memory devices, examples of which include semiconductor memory devices such as EPROM, EEPROM, and flash memory devices; magnetic disks such as internal hard disks or removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. Processors and memory can be supplemented by or incorporated into dedicated logic circuits.

[0073] To enable user interaction, embodiments of the subject matter described herein can be implemented using a computer equipped with a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) and a keyboard and pointing device (e.g., a mouse or trackball) that allows the user to input into the computer. User interaction can also be achieved using other types of devices; for example, the feedback provided to the user can be any form of sensory feedback, such as visual, auditory, or tactile feedback; and user input can be received in any form, such as acoustic, linguistic, or tactile input. Furthermore, the computer can also interact with the user by sending and receiving documents to and from the user's device; for example, this can be achieved by sending a web page to a web browser on the user's device in response to a request from a web browser.

[0074] Embodiments of the subject matter described herein may be implemented in a computing system that includes a backend component (e.g., a data server), a middleware component (e.g., an application server), a frontend component (e.g., a client computer with a graphical user interface or web browser that allows a user to interact with an implementation of the subject matter described herein), or any combination of one or more such backend, middleware, or frontend components. The components of the system can be interconnected by digital data communication in any form or medium, such as a communication network. Examples of communication networks include local area networks (LANs) and wide area networks (WANs), such as the Internet.

[0075] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically exchange information over a communication network. The client-server relationship arises from computer programs running on each computer that have a client-server relationship with one another.

[0076] This specification includes many specific implementation details, which should not be interpreted as limiting the scope of the claims, but rather as descriptions of features that may be specific to particular embodiments. Certain features described herein in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any preferred partial combination in multiple embodiments. Furthermore, features may be described above as functioning in a particular combination, and even if this is initially asserted, one or more features may be omitted from the asserted combination, or the asserted combination may refer to a partial combination or a variation of a partial combination.

[0077] Similarly, while the drawings show operations in a specific order, this should not be interpreted as meaning that such operations must be performed in a specific order or sequence shown, or that all illustrated operations must be performed, in order to obtain the desired results. Under certain circumstances, multitasking or parallel processing may be advantageous. Furthermore, the separation of various system components in the embodiments described above should not be interpreted as meaning that such separation is necessary in all embodiments, and it should be understood that the described program components and systems can generally be integrated into a single software product or packaged into multiple software products.

[0078] Specific embodiments of this subject matter have been described. Other embodiments are also included in the following claims. For example, the operations described in the claims may achieve the desired results even if they are performed in a different order. As an example, the process shown in the accompanying drawings does not necessarily require a specific order, i.e., sequence, to achieve the desired results. In certain embodiments, multitasking or parallel processing may be advantageous. Other steps or stages may be added to or omitted from the described process. Therefore, other embodiments are also included in the following claims.

[0079] Several embodiments In some embodiments, a method for conducting a cross-sectional or longitudinal human study or trial includes the step of separating data from a database of records of multiple individuals into input data and output data, where the input data are static or temporal features of each individual, collected either passively or through active involvement, and the output data are primary stratification axes representing each individual's transition from relative health status to relative dysfunction or disease. Furthermore, the method includes the step of selecting one or more secondary stratification axes (e.g., sex, age, demographics) derived from data elements collected from the database of records of individuals. The method further includes the step of training a model using data from individuals for which input data and output data (e.g., ground truth output information) exist or can be complemented along the stratification axes. The method further includes the step of expanding the model to generate localization or probability distribution embeddings for each individual along the stratification axes using input data related to each individual's health status, or information correlated with current or future health status. The method further includes a step of stratifying each individual who can be recruited for a cross-sectional or longitudinal human study or trial based on the relationship between their respective collected data and a selected axis. Furthermore, the method includes a step of determining the target distribution of patients to be recruited, depending on the patient's position or probability distribution along the selection axis, where the patient's position or probability along the selection axis is separate from the population distribution of patients. Furthermore, the method includes a step of selecting and outputting a subgroup of several individuals who can be recruited for a cross-sectional or longitudinal human study or trial, where the collective distribution of the position or probability of the subgroup along the selection axis is predicted to be similar to the target distribution.

[0080] In some embodiments, the method further includes the step of training a model using input data and ground truth output information for each individual along a stratification axis, or data from individuals that can complement such information.

[0081] In some embodiments, the method further includes the step of expanding the model to generate localizations or probability distribution embeddings for each individual along a stratification axis, using input data related to each individual's health status or information correlated with their current or future health status.

[0082] In some embodiments, the method includes the step of collecting data from multiple individuals.

[0083] In some embodiments, data of multiple individuals includes: demographics, medical interviews, physical examinations, diagnoses, medications, treatments, vital sign measurements, vaccinations, presentation of test results, collection and / or analysis of serum, urine or other biological samples, imaging, medical records, radiographic reports, genetic tests or reports, biomarkers, pathology reports, and other health-related documented information, social determinants of health, financial status or behavior, or consumer status or behavior.

[0084] In some embodiments, the selected stratification axis may be one or more primary stratification axes and / or may range from relative health status to relative disease status. In some embodiments, the stratification axis may include one or more secondary stratification axes and / or may be selected based on factors such as known or assumed relationships between the axis and the disease status of interest, or the availability of data elements corresponding to the axis.

[0085] In some embodiments, the method further includes the step of selecting a secondary stratification axis representing one or more characteristics of multiple individuals (e.g., characteristics such as physiological comorbidities, demographics, socioeconomic variables).

[0086] In some embodiments, human studies or trials are intended to study the progression of health status, validate biomarkers, validate panels of biomarkers, validate targets, or evaluate the efficacy or safety of pharmacological, digital, or other forms of interventions, treatments, or behavioral changes in a population or subpopulation. In some embodiments, studying the progression of health status may be performed by a fully automated system, a semi-automated system, or by a human after automated indicators representing progression have been generated.

[0087] In some embodiments, the selected subgroups exhibit enhanced characteristics of one or more of the following: reduced physiological heterogeneity, increased probability of outcome events occurring, reduced heterogeneity of progression rate, increased likelihood of responding to intervention, increased likelihood of observing specific physiological or pathological signs, or, more broadly, increased homogeneity regarding the expected endpoint.

[0088] In some embodiments, the stratification axis represents a continuous or ordinal dimension.

[0089] In some embodiments, the stratification axis includes a single data element or a combination of multiple data elements collected from multiple individuals.

[0090] In some embodiments, an individual's position along a stratification axis may be represented by a point, a point estimate, or a probability distribution.

[0091] In some embodiments, the method further includes the step of estimating the location of an individual missing one or more relevant data elements using an imputation technique, where the imputation technique may include multiple imputation or regression models.

[0092] In some embodiments, the stratification axis includes trained parameters derived from a model trained on data elements collected from individuals.

[0093] In some embodiments, the determined target distribution is selected based on characteristics relating to the biology, biomarkers, biomarker panel, therapeutic target, or pharmacological intervention being evaluated.

[0094] In some embodiments, the target distribution is uniform with respect to the stratification axis. In some embodiments, the method further includes the discovery or validation of biomarkers or biomarker panels. In some embodiments, the target distribution is enhanced with respect to one or more stratification axis regions or subregions. In some embodiments, the method further includes evaluating the effectiveness of the intervention. In some embodiments, the selected target distribution may differ substantially from the background distribution of the individual patients being recruited, which may necessitate further recruitment in an uninducible recruitment protocol that seeks to recruit patients constituting the target distribution.

[0095] In some embodiments, the output subgroup includes individuals whose distribution along the stratification axis is statistically similar to the target distribution.

[0096] In some embodiments, the selected subpopulation allows for rapid recruitment of individuals across dimensions that might otherwise require longitudinal studies. In some embodiments, the selected subpopulation allows for efficient recruitment of individuals within a desired window (e.g., a portion or subset of a dimension).

[0097] In some embodiments, a selected subgroup of individuals may be chosen to minimize or maximize a particular characteristic, for example, to minimize the total cost or number of patients required to maintain a minimum coverage rate across a continuous dimension.

[0098] In some embodiments, a selection of individuals may be chosen to maintain a particular distribution or balance of covariates, including demographic, covariates, comorbidities, drug use, medical history, or other socioeconomic or biological characteristics.

[0099] In some embodiments, patients are selected by optimization methods, such as linear programming, gradient descent, dynamic programming, or other schemes that approximate the identification of the optimal subset of recruits.

[0100] In some embodiments, the output of a subgroup may report an empty output subgroup, indicating that no subgroups satisfy the target distribution and selection criteria. In some embodiments, if an empty output subgroup is reported, the test designer is suggested to refine or relax the specific target distribution or selection criteria so that a non-empty output subgroup is returned.

[0101] In some embodiments, the method further includes the step of outputting a list of patients having equivalent characteristics with respect to a stratification axis, along with the number of patients to be recruited from each list to satisfy the target distribution.

[0102] In some embodiments, the output of a subgroup of individuals is performed using anonymization markers.

[0103] In some embodiments, certain diseases have clear biomarkers that indicate the disease or its stage. In other diseases, such as liver disease, imaging can reveal the amount of fat or the degree of fibrosis in the liver, and in the case of Parkinson's disease, current diagnostic criteria rely on subjectively measuring the patient's tremors or sleep disturbances. Therefore, stratification allows for the quantification of conditions that have previously been subjectively defined. Stratification can be performed using percentage groups or a range of continuous scores. Stratification may also be performed using discrete groups.

[0104] In some embodiments, the stratification may be risk stratification. In embodiments, this method can be used to provide a risk hierarchy for life insurance adjusters or other risk-based applications.

[0105] In some embodiments, the method can provide clinical decision support by providing physicians, nurses, nurse practitioners, physician assistants, or patients with risk classes derived from the probability distribution output by the model. In some embodiments, physicians, nurses, nurse practitioners, physician assistants, or patients can access the model via the internet to determine a diagnosis or disease stage. In some embodiments, the method may be provided to physicians, nurses, nurse practitioners, physician assistants, or patients as a Software-as-a-Service (SaaS) implementation. In some embodiments, the same probability distribution output by the model is used to simultaneously provide clinical decision support and clinical trial enhancement.

[0106] In some embodiments, the model is a machine learning model or a neural network. The method can further utilize supervised machine learning techniques that train on ordered labels. In some embodiments, the predictive model is trained through ordinal regression, neural networks, generative models, random forests, nearest neighbor search, support vector machines, and Gaussian processes. Those skilled in the art will understand that other statistical methods, machine learning techniques, and neural networks can be used.

[0107] In some embodiments, the progression of health status is represented by continuous or discrete dimensions.

[0108] In some embodiments, the method further includes identifying or validating biomarkers or biomarker panels that indicate progression in a health status, biological function, physiological function, disease, or a combination of conditions. In some embodiments, the biomarkers may be a pair of biomarkers that define pairing of biomarker interventions. The method can determine whether two biomarkers can pair when indicating a condition, or whether biomarkers can pair with an intervention. In some embodiments, the method further includes using a model to assess whether a relationship exists between the biomarkers or biomarker panels and the progression of a health status or disease. The model may be a machine learning model or a neural network. In some embodiments, the method further includes characterizing the relationship between the biomarkers or biomarker panels and the progression of a health status or disease. This relationship may include the probability that the biomarkers or biomarker group indicate a progression of a health status, the confidence interval for that probability, and the time range to progression based on the range of biomarker values. In some embodiments, the biomarkers or biomarker panels are validated in a human study involving participants who are members of a subpopulation identified by the model.

[0109] In some embodiments, the method further includes a step of determining whether any intervention is effective in a human study or trial using a subpopulation selected by the model. In some embodiments, any intervention may include a drug intervention, a behavioral intervention, a digital intervention, data collection, or other intention or other action. In some embodiments, a digital intervention may be a mobile or computer application (e.g., “app”) that provides access to an application for purposes such as behavioral change, information on medication adherence, input of health indicators, and recording of diet and exercise data. In some embodiments, the human study or trial includes a step of determining which stage of health status, biological function, physiological function, or disease a particular intervention is effective in. In several embodiments, any intervention may be a method of determining the correlation between the intervention and changes in an individual’s existing medical records and biomarkers.

[0110] In some embodiments, the method further includes enhancing a human study or trial by selecting a subgroup in a specific health status, biological function, physiological function, or stage of disease, so that the effects of the intervention during the trial can be more accurately determined.

[0111] In some embodiments, the method further includes enhancing a human study or trial by selecting a subgroup based on predicted progression so that the effects of the intervention during the trial can be determined.

[0112] In some embodiments, the predicted progression is a known or existing medically agreed stage.

[0113] In some embodiments, the method further includes using interventions directed by a subpopulation model and conducting human studies or trials to evaluate the effectiveness and safety of the protocol being applied.

[0114] In some embodiments, the model is configured to identify individuals within a subgroup who should receive intervention, and to determine when the identified individuals should receive the intervention, or what intervention dose or regimen should be used.

[0115] In some embodiments, anonymization markers are used for the output of subgroups of individuals.

[0116] In some embodiments, the model is a machine learning model or a neural network.

[0117] In some embodiments, the method further includes the step of enrolling a subgroup in a clinical trial based on stratification. In some embodiments, the method further includes the step of retraining a model based on the results of the clinical trial.

[0118] In some embodiments, the method further includes the step of treating a subpopulation with a stratified treatment. In some embodiments, the method further includes the step of retraining a model based on the results of a clinical trial.

[0119] In some embodiments, a method for training a machine learning model to conduct a cross-sectional or longitudinal human study or trial includes the step of separating data from a database of records of multiple individuals into input data and output data. The input data consists of static or temporal features of each individual, collected either passively or through active involvement. The output data represents each individual's transition from a healthy state to dysfunction or disease. The method includes the step of defining one or more stratification axes derived from one or more data elements collected from the database of records of multiple individuals. Furthermore, the method includes a step of training a model using data from individuals for which input and output data (e.g., ground truth output information) exist or can be supplemented along a stratification axis, wherein the trained model is deployed and configured to perform the following: (i) an operation to generate localization or probability distribution embeddings for each individual along a stratification axis using input data related to each individual's health status or information correlated with their current or future health status; (ii) an operation to stratify each individual who can be recruited for a cross-sectional or longitudinal human study or trial based on the relationship between their respective collected data and the selected axis; (iii) an operation to determine a target distribution of patients to be recruited according to the patient's location or probability distribution along the selection axis, wherein the patient's location or probability along the selection axis is separate from the population distribution of patients; and (iv) an operation to select and output a subset of several individuals who can be recruited for a cross-sectional or longitudinal human study or trial, wherein the collective distribution of the locations or probabilities along the selected axis is predicted to be similar to the target distribution.

[0120] In some embodiments, a method for running a model trained to select a target population for a cross-sectional or longitudinal human study or trial includes the step of deploying the model to generate localizations or probability distribution embeddings for each individual along a stratification axis, using input data related to each individual's health status, or information correlated with their current or future health status. The model is trained on a database of individual records separated into input and output data. The input data consists of static or temporal features of each individual, collected either passively or through active involvement. The output data represents each individual's transition from relative health status to relative dysfunction or disease. The stratification axis can be defined based on one or more data elements collected from a database of records of multiple individuals. The model is trained on data from individuals for which input and output data (e.g., ground truth output information) exist or can be supplemented for each individual along the stratification axis. Each individual who can be recruited for a cross-sectional or longitudinal human study or trial is stratified based on the relationship between their respective collected data and the selected axis. Furthermore, the method includes a step of determining a target distribution of recruited subjects defined by the objectives of the study or trial, based on their position or probability distribution along the selection axis, where the position or probability of subjects along the selection axis is separate from the population distribution of patients. The method further includes a step of selecting and outputting a subgroup of individuals who can be recruited for a cross-sectional or longitudinal human study or trial, wherein the collective distribution of the position or probability of the subgroups along the selection axis is predicted to be similar to the target distribution.

[0121] term The expressions and terms used herein are for illustrative purposes only and should not be construed as limiting.

[0122] As used herein and in the claims, the terms “approximately,” the phrase “approximately equal,” and other similar phrases (e.g., “X has approximately the value of Y” or “X is approximately equal to Y”) should be understood to mean that one value (X) is within a given range of the other value (Y). Unless otherwise indicated, the given range may be ±20%, 10%, 5%, 3%, 1%, 0.1%, or less than 0.1%.

[0123] Measurement values, sizes, quantities, etc., may be presented in range form in this specification. Descriptions in range form are for convenience and simplification only and should not be interpreted as strict limitations on the scope of the invention. Therefore, range descriptions should be considered to specifically disclose not only individual numerical values ​​within that range, but also all conceivable subranges. For example, a range description such as 10–20 inches should be considered to specifically disclose subranges such as 10–11 inches, 10–12 inches, 10–13 inches, 10–14 inches, 11–12 inches, and 11–13 inches.

[0124] The indefinite articles “a” and “an” used in this specification and claims should be understood to mean “at least one” unless explicitly stated otherwise. The phrases “and / or” used in this specification and claims should be understood to mean “either one or both” of the elements thus combined, that is, elements that exist sometimes conjunctly and sometimes separately. Similarly, any multiple elements listed using “and / or” should be interpreted as “one or more” of the elements thus combined. In addition, other elements may optionally exist besides those specifically identified by the “and / or” phrase, whether related to or unrelated to the specifically identified elements. Therefore, as a non-restrictive example, when the statement “A and / or B” is used with an open expression such as “inclusive,” it may refer to A only (optionally including elements other than B); B only (optionally including elements other than A); and both A and B (optionally including other elements) in yet another embodiment.

[0125] Where used herein and in the claims, “or” should be understood to have the same meaning as “and / or” as defined above. For example, when separating items in a list, “or” or “and / or” is interpreted as inclusive, meaning that it includes at least one of a certain number of elements or lists, but also two or more, and optionally any additional unlisted items. Only terms that clearly indicate the opposite meaning (e.g., “either one,” “strictly one,” etc.) or, where used in the claims, terms such as “consisting of,” refer to including exactly one element from a number of elements or list. In general, where used, “or” should be interpreted as indicating an exclusive choice (i.e., “either one or the other, but not both”) only when preceded by terms indicating exclusivity such as “either,” “one of,” “only one of,” or “strictly one of.” Where used in the claims, “essentially consisting of” should have the usual meaning as used in the field of patent law.

[0126] As used herein and in the claims, the phrase “at least one” with respect to a list of one or more elements should be understood to mean at least one element selected from any one or more elements contained in the list of elements, but not necessarily including at least one of all elements specifically enumerated in the list of elements, nor excluding any combination of elements in the list of elements. This definition also allows for the presence of elements other than those specifically identified in the list of elements indicated by the phrase “at least one,” whether or not they are related to those specifically identified elements. Therefore, as a non-restrictive example, "at least one of A and B" (or synonymously, "at least one of A or B," or similarly, "at least one of A and / or B") may refer to, in one embodiment, no B (optionally including elements other than B) and at least one (optionally two or more) A; in another embodiment, no A (optionally including elements other than A) and at least one (optionally two or more) B; and in yet another embodiment, at least one (optionally two or more) A and at least one (optionally two or more) B (and optionally including other elements); and so on.

[0127] The use of “include,” “constitute,” “have,” “contain,” “include,” and variations thereof is intended to encompass the items and additional items subsequently listed.

[0128] The use of ordinal terms such as “first,” “second,” and “third” to modify claim elements in the claims does not, in itself, imply that one claim element has priority, precedence, or order over another, nor does it imply a chronological order in which the actions of the method are performed. Ordinal terms are merely used as identifiers to distinguish one claim element having a particular name from another element having the same name (except for the use of ordinal terms).

[0129] While specific examples of embodiments have been shown and described, those skilled in the art will understand that various modifications can be made in form and detail without departing from the scope of embodiments covered by the appended claims.

Claims

1. A method for conducting clinical trials or biomarker validation studies: A step of obtaining access to a computer implementation model trained with training data containing multiple records relating to multiple individuals, wherein each record includes (i) values ​​of multiple features relating to each individual obtained from the multiple individuals, and (ii) a label indicating the pathological condition of each individual; A step of providing a trained computer implementation model with multiple sets of values ​​for the multiple features, wherein each set of values ​​corresponds to a candidate individual from a set of candidate individuals; The steps include: receiving a prediction of the disease state for each set of values ​​from the aforementioned trained computer implementation model; The steps include: identifying the participant group from the set of candidate individuals based on the prediction of the aforementioned pathological condition; A method comprising the step of promoting at least one clinical trial or biomarker validation study in which the aforementioned group of participants participates.

2. The method according to claim 1, wherein the values ​​of the plurality of features describe at least one of demographics, medical interviews, physical examinations, diagnoses, drugs, medical procedures, vital sign measurements, vaccinations, test results, serum, urine samples, biological samples, gene expression levels, medical images, medical records, radiographic reports, genetic tests, biomarkers, pathology reports, health information, social determinants of health, financial data, consumer data, or any combination thereof.

3. The method according to claim 1, wherein the model includes at least one of a regression model, a classifier, a linear model, a nonlinear model, a random forest, a kernel method, a Bayesian model, a decision tree, or a neural network.

4. The method according to claim 1, wherein the prediction of the disease state includes a probability distribution.

5. The method according to claim 1, wherein the prediction of the disease state is based on a primary stratification axis including a health-to-disease axis or a disease risk axis.

6. The method according to claim 1, wherein the identification of the participant group includes stratifying each of the multiple candidate individuals from the multiple candidate individuals according to one or more primary stratification axes.

7. The method according to claim 6, wherein the identification of the participant group further comprises stratifying each of the candidate individuals from the plurality of candidate individuals according to one or more secondary stratification axes, wherein the one or more secondary stratification axes represent one or more characteristics of the plurality of individuals, the characteristics include at least one of physiological comorbidities, demographics, or socioeconomic variables.

8. The method according to claim 7, wherein at least one stratification axis from the one or more primary stratification axes or the one or more secondary stratification axes represents at least one continuous dimension or sequential dimension.

9. The method according to claim 7, wherein the positions along at least one stratification axis from the one or more primary stratification axes or the one or more secondary stratification axes are represented by points, point estimates, or probability distributions.

10. The method according to claim 1, wherein the clinical trial or biomarker validation study includes studying at least one of the following: the progression of health status changes, the validity of biomarkers, the validity of a biomarker panel, the validity of a targeted therapy, the effectiveness of a therapeutic intervention, the effectiveness of a digital intervention, the effectiveness of a behavioral intervention, or any combination thereof.

11. The method according to claim 1, wherein the participant group is enhanced in one or more traits compared to the set of candidate individuals, the one or more traits comprising at least one of the following: reduced physiological heterogeneity, increased probability of outcome event occurrence, increased tendency to respond to intervention, increased likelihood of observing specific physiological or pathological signs, or any combination thereof.

12. The identification of the aforementioned participant group is: A step of determining a target distribution for the set of candidate individuals in accordance with the objectives of the clinical trial or the biomarker validation study, wherein the target distribution defines a probability distribution; The steps include selecting the participant group from a set of candidate individuals so as to achieve a collective distribution of the participant group that is similar to the target distribution, The method according to claim 1, including the method described in claim 1.

13. The method according to claim 12, wherein the target distribution is determined based on at least one parameter to be evaluated in the clinical trial or the biomarker validation study, the at least one parameter comprising at least one of a biological characteristic, a biomarker, a biomarker panel, a therapeutic target, or a pharmacological intervention.

14. The method according to claim 12, wherein the target distribution is uniform with respect to the pathological condition.

15. The method according to claim 12, wherein the target distribution includes a risk probability distribution.

16. The method according to claim 12, wherein the target distribution is enhanced with respect to the region of the stratification axis related to the pathological condition.

17. The method according to claim 12, wherein the step of selecting the participant group includes minimizing the number of individuals within the participant group.

18. The selection of the aforementioned participant group is: A step of identifying multiple groups of individuals, wherein each of the multiple groups of individuals has a similar probability distribution with respect to the disease state; The steps include determining the number of individuals from each group of individuals to be included in the participant group in order to satisfy the target distribution, The method according to claim 12, including the method described in claim 12.

19. The method according to claim 1, wherein the method comprises accelerating a clinical trial, and the clinical trial comprises evaluating the effectiveness of the treatment.

20. The method according to claim 1, wherein the method comprises facilitating a clinical trial, the facilitating the conduct of the clinical trial comprising enrolling members from the participant group in the clinical trial, thereby providing the members with a drug, a digital intervention, or a behavioral intervention.

21. The method according to claim 1, wherein the method comprises facilitating the biomarker validation study, the biomarker validation study comprising validating a biomarker or a biomarker panel.

22. The method according to claim 1, wherein the step of identifying the participant group comprises achieving a desired distribution of at least one covariate, the at least one covariate including demographics, comorbidities, drug use, medical history, socioeconomic characteristics, or biological characteristics.

23. The method according to claim 1, further comprising the step of training a computer implementation model using the training data.

24. A system for conducting clinical trials or biomarker validation studies, wherein the system: To obtain access to a computer implementation model trained with training data containing multiple records relating to multiple individuals, wherein each record contains (i) multiple feature values ​​relating to each of the multiple individuals, and (ii) a label indicating the pathological condition of each individual; To the trained computer implementation model, each set of multiple values ​​corresponding to a candidate individual from a set of candidate individuals is provided to the trained computer implementation model; The trained computer implementation model provides a prediction of the pathological condition for each set of values; To identify a group of participants from the set of candidate individuals based on the prediction of the aforementioned pathological conditions; and To promote at least one clinical trial or biomarker validation study in which the aforementioned participant group participates. A system including one or more computer processing units programmed to perform operations including the following.

25. The system according to claim 24, wherein the prediction of the disease state is based on a primary stratification axis including a health-to-disease axis or a disease risk axis.

26. The system according to claim 24, wherein the identification of the participant group includes the step of stratifying each candidate from the plurality of candidate individuals according to one or more primary stratification axes and one or more secondary stratification axes.

27. The identification of the aforementioned participant group is: A step of determining a target distribution for the set of candidate individuals in accordance with the objectives of a clinical trial or biomarker validation study, wherein the target distribution defines a probability distribution; The steps include selecting the participant group from the set of candidate individuals in order to achieve a collective distribution of the participant group that is similar to the target distribution, The system according to claim 24, including the system described in claim 24.

28. The system according to claim 27, wherein the target distribution includes a risk probability distribution.

29. The selection of the aforementioned participant group is: A step of identifying multiple groups of individuals, wherein each of the multiple groups of individuals has a similar probability distribution with respect to the disease state; The steps include determining the number of individuals from each group of individuals to be included in the participant group in order to satisfy the target distribution, The system according to claim 27, including the system described in claim 27.

30. The system according to claim 24, wherein the operation includes training a computer-implemented model using the training data.

Images