System and method for digital twin-based metabolic health optimization using longitudinal biomarker and digital resilience analytics

Abstract

A digital twin for improving physiological resilience of a user. The digital twin uses longitudinally collected user data comprising a health regimen of the user and, from a plurality of sensors, measurements indicative of biomarker levels of the user during performance of the health regimen. In particular, at least one resilience biomarker is computed from the user data, and the digital twin generates a health intervention based on the user data and the at least one resilience biomarker, the health intervention being selected to achieve a predetermined change in the at least one resilience biomarker. The digital twin also predicts a predicted change in the at least one resilience biomarker corresponding to adherence to the health intervention.

Description

[0001] System and Method for Digital Twin-Based Metabolic Health Optimization Using Longitudinal Biomarker and Digital Resilience Analytics

[0002] Technical Field

[0003] The present invention relates to systems and methods for providing and using a digital twin for metabolic health optimization based on longitudinal monitoring of metabolic flexibility score (MFS). The invention further relates to assessing the impact of health optimization and behaviour changes from functional digital resilience biomarkers.

[0004] Background

[0005] This background is provided for context and understanding of the disclosure. Contents of this background section are neither expressly nor implied admitted as prior art against the present disclosure.

[0006] Cardiometabolic diseases (CMDs), including metabolic syndrome (MetS), type 2 diabetes mellitus (T2DM), and cardiovascular diseases (CVDs), remain major global health burdens primarily influenced by sedentary lifestyles, poor diets, and aging. MetS is characterized by elevated blood glucose, hypertension, and obesity, which, if left unmanaged, can lead to T2DM and CVD. Although metabolic disorders are generally more prevalent with age, their onset is increasingly observed among younger individuals due to high-calorie diets, sedentary behaviour, and other socioeconomic factors.

[0007] Early-stage metabolic abnormalities can often be reversed through timely lifestyle modification and medication. However, major challenges in disease prevention include low adherence to healthy routines and the lack of tools for early detection of asymptomatic metabolic decline. These challenges highlight the need for effective, personalized methods that facilitate sustainedbehavioural change and enable early assessment of metabolic health across diverse populations.

[0008] Metabolic flexibility and metabolic switching refer to the body's ability to switch between glucose and fat utilization for energy production, is an important indicator of metabolic health, reflecting the body's adaptability to physiological challenges. While professional athletes typically exhibit high metabolic flexibility, interindividual variation also exists among non-athletes due to differences in genetics, health status, and fitness levels. Not all obese individuals are metabolically inflexible, and some exhibit transcriptional profiles similar to normal-weight individuals, suggesting that metabolic flexibility can serve as an early biomarker of metabolic resilience independent of body size.

[0009] Recent advances in wearable technologies, continuous monitoring sensors, and digital health platforms have enabled individuals to longitudinally track physiological and biochemical parameters relevant to metabolic and overall health. By analysing such time-series data, digital biomarkers can be derived to capture dynamic physiologic responses to interventions, including diet, exercise, sleep, and stress management. These derived digital resilience biomarkers provide information on the adaptability and recovery capacity of the body, features that static measurements often fail to reveal.

[0010] As the diversity and resolution of biomarker data increase, correlations between biological markers (e.g., blood-based analytes) and digital signals (e.g., wearable-derived metrics) may allow for inference of one from the other. Defining and validating these digital biomarkers can facilitate personalized intervention strategies, improve user engagement, and enable early detection of metabolic or aging-related dysfunctions. Moreover, such longitudinal tracking approaches are expected to accelerate the development of next-generation continuous sensing devices for other biochemical markers, similar to continuous glucose monitors.Accordingly, there remains a need for systems and methods that integrate longitudinal biomarker monitoring with digital resilience analytics to evaluate and optimize metabolic health, quantify the effects of behavioural and lifestyle interventions, and promote sustainable health improvements.

[0011] Summary

[0012] Disclosed is a computer-implemented method for providing a digital twin for improving physiological resilience of a user, comprising steps of:

[0013] receiving longitudinally collected user data comprising a health regimen of the user and, from a plurality of sensors, measurements indicative of biomarker levels of the user during performance of the health regimen;

[0014] computing at least one resilience biomarker based on the user data; and generating the digital twin for the user, the digital twin being configured to generate:

[0015] a health intervention based on the user data and the at least one resilience biomarker, the health intervention being selected to achieve a predetermined change in the at least one resilience biomarker; and a predicted dynamic change in the at least one resilience biomarker corresponding to adherence to the health intervention.

[0016] The computer-implemented method may further comprise:

[0017] periodically receiving updated user data comprising adherence to the health intervention and further measurements of the biomarker levels from the plurality of sensors; and

[0018] updating the digital twin, thereby to update the health intervention, based on the updated user data.

[0019] Also disclosed is a system for providing a digital twin for improving physiological resilience of a user, comprising:

[0020] memory;

[0021] at least one processor;at least two sensors; and

[0022] the digital twin,

[0023] the memory storing instructions that, when executed by the at least one processor, cause the system to:

[0024] receive longitudinally collected user data comprising a health regimen of the user and, from a plurality of sensors, measurements indicative of biomarker levels of the user during performance of the health regimen;

[0025] compute at least one resilience biomarker based on the user data; and generate the digital twin for the user, the digital twin being configured to generate:

[0026] a health intervention based on the user data and the at least one resilience biomarker, the health intervention being selected to achieve a predetermined change in the at least one resilience biomarker; and a predicted dynamic change in the at least one resilience biomarker corresponding to adherence to the health intervention.

[0027] According to a further aspect of the present invention, there is provided a computer-implemented method for determining a global resilience index test (GRIT) composite physiological resilience score, comprising:

[0028] receiving, from one or more sensors or data sources, biomarker data comprising longitudinal measurements of a plurality of biomarkers of a user taken during performance of a health regimen, stressor, or intervention;

[0029] generating for each said biomarker or groups of said biomarkers, a respective resilience score based on a piecewise scoring function defined for the corresponding biomarker; and

[0030] aggregating the resilience scores to compute the GRIT score, the GRIT score representing an overall physiological resilience of the user.

[0031] The computer-implemented method may further comprise determining a GRIT biological age based on the GRIT score and a chronological age of the user.Brief description of the drawings

[0032] Embodiments of the present invention will now be better understood and readily apparent to one of ordinary skill in the art from the following written description, by way of example only, and in conjunction with the drawings, in which:

[0033] Figure 1 depicts a schematic flow diagram of a method for providing a digital twin for metabolic health optimization based on longitudinal monitoring resilience biomarkers, particularly metabolic flexibility score (MFS), of a user.

[0034] Figure 2 illustrates a diagram of a workflow to compute resilience biomarkers, according to various embodiments of the present invention.

[0035] Figure 3 illustrates a diagram of a workflow to generate the digital twin model, according to various embodiments of the present invention.

[0036] Figure 4 is a schematic overview of optimizing metabolic health with digital twins, according to various embodiments of the present invention.

[0037] Figure 5 depicts overview of the real-world health intervention trial for subject N001 who adheres to a health regimen and collects biomarker responses longitudinally, according to various embodiments of the present invention.

[0038] Figure 6 depicts a schematic flow diagram of a method for computing an aggregated resilience score representing an overall physiological resilience of the user, according to various embodiments of the present invention.

[0039] Figure 7 depicts graphs for glucose-ketone dynamics during a strengthcentric workout, according to various embodiments of the present invention.

[0040] Figure 8 depicts a graph for homocysteine dynamics during a fasting-refeed cycle, according to various embodiments of the present invention.Figure 9a depicts a graph for metabolic flexibility score (MFS) providing a scoring system based on a the time to ketosis, according to various embodiments of the present invention.

[0041] Figure 9b depicts a graph for metabolic flexibility score (MFS) providing a scoring system based on ketone / glucose trajectories, according to various embodiments of the present invention.

[0042] Figure 10 depicts graphs for glucose-ketone dynamics during the one-meal-a-day (OMAD) ketosis entry regimen, according to various embodiments of the present invention.

[0043] Figure 11 depicts a graph for base score for the methylation resilience score, according to various embodiments of the present invention.

[0044] Figure 12 depicts graphs for base score for lipid regulation efficiency, according to various embodiments of the present invention.

[0045] Figure 13 depicts graphs for inflammation score for measured hsCRP levels, according to various embodiments of the present invention.

[0046] Figure 14 depicts graphs for insulin sensitivity score for measured HOMA-IR levels, according to various embodiments of the present invention.

[0047] Figure 15 depicts graphs for GRIT Biological Age, according to various embodiments of the present invention.

[0048] Detailed description

[0049] It will be appreciated that many further modifications and permutations of various aspects of the described embodiments are possible. Accordingly, thedescribed aspects are intended to embrace all such alterations, modifications, and variations that fall within the spirit and scope of the appended claims.

[0050] Throughout this specification and the claims which follow, unless the context requires otherwise, the word "comprise", and variations such as "comprises" and "comprising", will be understood to imply the inclusion of a stated integer or step or group of integers or steps but not the exclusion of any other integer or step or group of integers or steps.

[0051] Public Health Burden and Challenges in Managing Cardiometabolic Diseases

[0052] Cardiometabolic diseases (CMDs), encompassing metabolic syndrome (MetS), type 2 diabetes mellitus (T2DM), and cardiovascular diseases (CVDs), represent a major global health burden. Aging, poor dietary habits, and sedentary lifestyles are well-established risk factors that contribute to the development and progression of these disorders. Metabolic syndrome is characterized by a cluster of metabolic abnormalities, including central obesity, hypertension, dyslipidemia, and hyperglycemia, which collectively elevate the risk of T2DM and CVDs if not effectively managed.

[0053] Early-stage metabolic disorders are often reversible through lifestyle modification and, in some cases, medication. However, low adherence to healthy behaviours and the lack of early detection of asymptomatic metabolic abnormalities remain key challenges. These issues underscore the need for effective, personalized methods to promote sustained lifestyle adherence and enable early detection of metabolic decline, including among generally healthy individuals, in decentralized settings.

[0054] Optimizing metabolic health with a metabolic flexibility-based digital twin

[0055] Professional athletes and individuals with metabolic syndrome (MetS) represent opposite ends of the metabolic health spectrum. Athletes exhibit high metabolicflexibility, the ability to efficiently switch between glucose and fat utilization in response to physiological demands, while individuals with MetS or aging-related decline show reduced adaptability to nutritional or energetic stress. Even among metabolically "healthy" non-athletes, interindividual variability in metabolic flexibility exists, influenced by genetic and lifestyle factors. Differences in fuelswitching capacity among healthy individuals can predict future metabolic risk, and some obese individuals retain transcriptional profiles associated with preserved flexibility. Accordingly, metabolic flexibility serves as an early biomarker of metabolic health, useful for stratifying risk and monitoring the evolving impact of lifestyle interventions over time.

[0056] The present disclosure introduces digital twin technology for managing metabolic resilience, the digital twin being an evolving digital entity designed to accurately reflect resilience biomarkers of an individual.

[0057] Disclosed is a digital twin model to monitor metabolic flexibility. The digital twin is configured to facilitate long-term adherence to lifestyle modifications by providing a guideline as to how resilience biomarkers change in response to adherence to a prescribed health intervention or health regimen. The digital twin thereby enables personalized health management, and supports early detection of subclinical metabolic decline. Embodiments of the proposed digital twin comprise two principal modules: (1) a gamification module configured to monitor longitudinal trends in fuel switching and metabolic flexibility, and to provide real-time dynamic feedback mechanisms based on resilience scores computed from biomarker readouts that can be modulated and / or respond to physiological stressors or interventions, such as prolonged fasting or exercise; and (2) a machine learning (ML) or artificial intelligence (AI)-powered integrated analysis module configured to analyse longitudinal biomarker data and predict long-term health outcomes associated with adherence to specific health regimens and metabolic flexibility trajectories.In preferred embodiments, the invention demonstrates the technical feasibility of constructing the digital twin model, identifies implementation challenges, and outlines the associated socioeconomic considerations required to enable scalable population-level improvements in metabolic health.

[0058] Gamification module of the digital twin - promoting adherence to health regimen through fuel switching monitoring

[0059] The present digital twin is a data-driven approach to assessing metabolic flexibility by tracking the dynamic relationship between blood glucose and ketone body (KB) levels during lifestyle or health interventions. This approach forms the foundation of a digital twin solution designed to enable decentralized metabolic health management and continuous evaluation of physiological adaptability under real-world conditions.

[0060] Under conditions of glycogen depletion, adipose tissue mobilizes free fatty acids that are subsequently converted into KBs such as β-hydroxybutyrate, acetoacetate, and acetone, which serve as alternative energy substrates in place of glucose. KBs have been extensively recognized as biomarkers of elevated fat metabolism, and are clinically relevant for predicting metabolic disease risk, detecting diabetic ketoacidosis (DKA), evaluating dietary interventions, and monitoring metabolic optimization.

[0061] Figure 1 illustrates a computer-implemented method 100 for providing a digital twin for resilience biomarker modelling for use, for example, in metabolic health optimization based on longitudinal monitoring of MFS and cardiometabolic health and aging biomarkers of a user. It begins with input data such as glucose and KB measurements and details of a fixed health regimen, followed by pattern analysis to identify the occurrence of fuel switching within a preset period. A Metabolic Flexibility Score (MFS) or other resilience biomarker(s) will be computed based on the fuel switching speed or other factors. For instance, an MFS of 90% is assigned if fuel switching occurs rapidly (e.g., day 2),encouraging adherence to the current regimen. If fuel switching occurs with delay (e.g., day 10), an MFS of 60% is given, recommending adherence and continued monitoring. If fuel switching does not occur within 14 days, it will prompt potential risk of metabolic disorder and immediate adjustments to the health regimen (e.g. reduce carbohydrate intake). The output provides actionable scores to gamify adherence, motivating individuals to optimize metabolic flexibility and healthy behaviour.

[0062] A gamification module facilitates visualisation of changes to resiliency biomarkers and / or other biomarkers, in response to performance of a health regimen, stressor, or intervention. In some embodiments, the gamification module takes feedback of biomarker levels and / or other user inputs, real-time or otherwise, for users to track their progress against predicted dynamic changes (which can also be referred to as a "predicted change" or similar, as dictated by context) to biomarker or resilience biomarkers levels assuming adherence to the health intervention. This reinforces motivation and adherence to the prescribed health regimen. In practical applications, the ability to observe successful fuel switching (e.g., transition into ketosis) encouraged sustained behavioural compliance, even under variable lifestyle conditions such as travel.

[0063] As shown, the method 100 begins at step 102, in which user data are received. The user data is collected through longitudinally (i.e., over time, such that changes in biomarker levels can be observed) monitoring biomarker levels during performance of a health regimen, stressor, or intervention, which can involve periods of particular activities (cardiovascular activity, weight lifting, anaerobic exercise and others), rest periods, fasting, eating and other phases. Usually, multiple biomarkers are monitored, though one biomarker may be monitored in some embodiments. In some embodiments, the sensor-based measurements are indicative of the user's metabolic flexibility, as well as measurements of cardiometabolic health and aging biomarkers. The metabolic flexibility measurements may include, for example, dynamic changes in bloodglucose and ketone levels, or other metabolic indicators derived from continuous or periodic monitoring devices.

[0064] The sensors can include wearable continuous glucose monitors, heart rate monitors, weighing scales, sphygmomanometers and non-wearable blood or tissue testing or sampling devices for periodic assessment of glucose and KB concentrations. Characteristic trajectories of glucose and KB dynamics were observed following sustained implementation of fitness and dietary interventions, reflecting the onset of ketosis and the degree of metabolic flexibility. These observations support the technical feasibility of quantifying metabolic flexibility through non-invasive, longitudinal biomarker tracking using data acquired from multiple sensor modalities.

[0065] At step 104, at least one resilience biomarker is computed based on the user data. For present purposes, the resilience biomarker or biomarkers will be represented by MFS, though it will be understood to apply similarly to other resilience biomarkers and to multiple resilience biomarkers. The MFS reflects the user's ability to switch between energy substrates such as glucose and fat in response to changes in physiological state or intervention, and serves as a dynamic indicator of metabolic adaptability and resilience.

[0066] Thereafter, at step 106, a digital twin for the user is then generated based on the MFS and user data. The digital twin is configured to generate a personalized health intervention based on the MFS and user data, the intervention being selected to achieve a predetermined change in one or more of the MFS, cardiometabolic health, or aging biomarkers. In particular, where the digital twin has been trained using biomarker data from a plurality of users, for a variety of health interventions, the digital twin may determine which health intervention is likely to yield greatest improvement in the resilience biomarker (e.g., greatest reduction in switching time for metabolic flexibility), and recommend that health intervention. This can be achieved by the digital twin predicting the dynamic change in resilience biomarker for each of a plurality of health interventions,and prescribing the health intervention that causes the greatest change in the resilience biomarker. The digital twin can thereby generate a health intervention and predict the dynamic change in the resilience biomarker in response to adherence to the health intervention.

[0067] As used in this context, the term "dynamic change" includes a specific figure a resilience biomarker is expected to reach at the end of a health intervention, at some point during a health intervention, or at some point after the long-term effects of the health intervention (e.g., sustained improved metabolic switching, that persists even when the person is not active for a short period, such as a week or other period specified by a clinician), and also includes a progressive change overtime, if the health intervention is adhered to.

[0068] In certain embodiments, the digital twin is configured to periodically receive updated user data for updating the digital twin and health intervention. The updated user data can include information relating to adherence to the health intervention and further measurements of biomarker levels from the sensors.

[0069] The module 1 of the proposed digital twin is designed to monitor fuel switching and promote adherence to healthy behaviours as shown in Figure 2, illustrating the workflow for the gamification module within the proposed digital twin that monitors fuel switching to compute a MFS. The score gamifies adherence to healthy behaviours by evaluating fuel switching speed and providing real-time dynamic feedback as the reward mechanism.

[0070] In some embodiments, the method further comprises generating a visualisation fora graphical user interface (GUI), showing the digital twin. The graphical user interface provides real-time feedback indicative of the predicted changes in the resilience biomarker. For example, the GUI can include predicted longitudinal trends of the MFS, cardiometabolic health, and aging biomarkers, based on the user's adherence to the health intervention. This enables the user to interactively track progress of measured biomarker levels or resiliencebiomarker(s) calculated therefore, against the predicted levels, and observe the effect of behavioural modifications. The digital twin can also apply predictive analytics for evaluation of long-term health outcome resulted from sustained fuel switching. For example, after predicting biomarker levels, and thus resilience biomarker levels, achievable in response to adherence to a prescribed health regimen for a predetermined period of time (e.g., three months), the digital twin may use the predicted changes (e.g., where those predictions relate to an outcome resilience biomarker or biomarker level, rather than a progressive change in that level) to determine a new baseline set of biomarker levels (i.e., the predicted biomarker levels at the end of the three month period of adherence to the health regimen) and generate a new health intervention and predicted change to biomarker and / or resilience biomarker levels. This process can be performed iteratively, so that a user can visualise the long-term benefit of continued adherence.

[0071] Figure 3 illustrates a workflow to generate the digital twin model. This workflow generates or builds a digital twin model that integrates health regimen adherence, baseline and post-intervention health data collection, and metabolic flexibility assessment. The artificial intelligence (AI)-powered module links health regimens with long-term health outcomes to provide predictive insights and personalized health interventions.

[0072] In another embodiment, the user data further comprises user health metrics and clinical biomarker data. The user health metrics may include lifestyle parameters such as sleep duration, weight, heart rate variability, and physical activity levels obtained from wearable devices, while the clinical biomarker data may be derived from laboratory analyses of biological samples such as blood, saliva, or urine. Integration of these multimodal data sources provides a comprehensive and dynamically evolving representation of the user's metabolic and physiological state within the digital twin framework.In some embodiments, the system may be configured to perform frequent or periodic post-intervention checkpoints comprising intermittent measurements of one or more biomarkers or biomarker sets selected from the group consisting of: (a) anthropometric markers such as body weight, body mass index (BMI), blood pressure, and body fat composition; (b) cardiometabolic biomarkers such as such as glucose, ketone, homocysteine, Apolipoprotein A (ApoA), Apolipoprotein B (ApoB), Hemoglobin A1c (HbA1C), and blood lipids; (c) inflammatory biomarkers such as Interleukin-6 (IL-6) and C-reactive Protein (CRP); (d) genetic biomarkers such as methylation patterns, telomere length, and genetic profile; and (e) microbiome biomarkers including gut, skin, oral, or vaginal microbiota. Baseline and follow-up measurements of the biomarkers may be used to assess longitudinal changes over time, which form part of the predicted change generated by the digital twin. The digital twin may further be configured to model the relationship between fuel switching occurrence, longitudinal trends of MFS, and variations in health metrics and biomarkers using Al-powered models.

[0073] In one embodiment, the digital twin further comprises an artificial intelligence (AI)-powered predictive analytics model trained to learn relationships between longitudinal changes in the biomarker levels during performance of the health regimen, and output a predetermined health regimen. To achieve this, the model uses baseline measurements of the biomarker levels for a plurality of users, longitudinal measurements of the biomarker levels during performance of at least one predetermined health regimen by those individuals (or one or more of a variety of health regimens, performed by respective individuals), to learn relationships between longitudinal changes in the biomarker levels during performance of the health regimen. The output is thus calculated from the longitudinal changes in the biomarker levels.

[0074] The Al-model of the digital twin can simulate or predict long-term health outcomes based on implemented interventions. Also, where relationships between biomarker are correlated, e.g., resting heart rate and frequency ofcardiovascular exercise, one biomarker may be used as a proxy for another biomarker. The digital twin may learn relationships between biomarkers - e.g., from paired data of both biomarkers or three or more biomarkers - such that data for one biomarker can be used to generate synthetic data (i.e., a prediction) corresponding to the other biomarker. The digital twin can therefore generate synthetic data for unmeasured biomarkers to aid in health trajectory forecasting and minimizing the need for invasive testing. Accordingly, one or more biomarkers used in computing the digital twin need not be obtained directly from sensor measurements, but may instead be derived, inferred, or computationally generated based on algorithmic relationships with other biomarkers.

[0075] For instance, a machine learning model may infer a user's ketone level from measured glucose data or predict other metabolic indicators from correlated parameters. In this way, certain biomarkers (such as a glucose-ketone index (GKI), homocysteine level, apolipoprotein B / apolipoprotein A (ApoB / ApoA) ratio, high-sensitivity C-reactive protein (hsCRP), or homeostatic model assessment of insulin resistance (HOMA-IR)) may be either directly measured or inferred from available data. This capability enables the digital twin to maintain continuity of physiological modelling even when some data streams are unavailable, thereby improving robustness of health trajectory forecasting and reducing reliance on invasive testing.

[0076] In addition, the Al-powered model may comprise a large language model (LLMs) trained to generate verbal or textual output to encourage adherence. The AI-powered model may also generate alerts if the user fails to adhere to a particular component of the health regimen - e.g., if a cardiovascular routine should start at a predetermined time, and biomarker measurements do not correspond to cardiovascular activity.

[0077] An example user interface of the digital twin includes longitudinal trends of MFS and predicted long-term biomarker changes as shown in Figure 4. As shown,achieving the predetermined change in the metabolic flexibility score (MFS) is indicative of long-term adherence to and effectiveness of the health intervention. The MFS serves as a quantifiable indicator of metabolic adaptability, and sustained improvement in the MFS may reflect enhanced metabolic efficiency and resilience, thereby validating the user's adherence and response to the prescribed regimen.

[0078] According to a preferred embodiment, Figure 4 illustrates a user interface of digital twins. In the left panel, a subject with declining MFS over a period of few months receives poor digital twin predicted outcomes for an array of biomarkers. In contrast, in the right panel the digital twin predicts optimal relevant biomarker results for a physically active subject with excellent MFS. Abbreviations: metabolic flexibility score (MFS); apolipoprotein A (ApoA); apolipoprotein B (ApoB); hemoglobin A1C (HbA1C); interleukin-6 (IL-6); C-reactive protein (CRP).

[0079] Trackinq biomarker and metabolic pathway requlatory chanqes in response to varied health interventions

[0080] In certain embodiments, the present invention provides a system and method for assessing and optimizing metabolic health by tracking the effects of behavioural and lifestyle interventions through functional digital resilience biomarkers. The system is configured to collect, integrate, and analyse longitudinal data comprising cardiometabolic, aging, and microbiome biomarkers obtained during adherence to one or more health regimens.

[0081] The collected data may include temporally resolved biomarkers, which represent snapshots of biomarker readings captured over time, for example, in response to a physiological stressor or intervention, and which are tracked longitudinally to enable comparison across multiple time points. By analysing variations in these biomarkers and associated regulatory pathways in response to interventions, the system provides adaptive feedback to the user or clinician,enabling quantifiable assessment of physiological changes and adherence effectiveness.

[0082] Digital resilience biomarkers may be derived from longitudinal trends, rates of change, or correlations among biomarkers, indicating an individual's adaptability to physiological and behavioural stressors. In some embodiments, the biomarkers are derived from glucose and ketone dynamics reflecting transitions between glycolytic and lipolytic fuel utilization states. These dynamics may be correlated with intervention parameters such as fasting, exercise, dietary modification, or sleep optimization to determine individualized metabolic switching rates.

[0083] In some embodiments, the system integrates molecular and wearable data sources, and the method is implemented in or integral with a wearable device. Physiological signals such as heart rate variability, sleep efficiency, and activity level may be analysed together with blood or microbiome-based biomarkers. Algorithmic or machine learning models may infer unmeasured parameters, such as using one or more biomarkers as a proxy for other biomarkers, thereby enhancing metabolic monitoring with minimal invasive testing.

[0084] Resilience parameters computed by the system may include metabolic switching latency, glycolytic onset rate, ketotic recovery rate, circadian stability indices, or recovery dynamics following physiological challenges. These parameters may be computed using predictive or model-based analytics applied to longitudinal datasets, providing personalized indices of metabolic adaptability beyond static biomarker values. The system may further incorporate continuous or near-continuous biosensors for tracking biomarkers beyond glucose and ketone, such as C-reactive protein (CRP), homocysteine, or other relevant analytes, individually or in combination, to improve predictive health modelling.

[0085] An experimental study demonstrated the feasibility of this framework in monitoring metabolic flexibility and reinforcing adherence to health regimens.Prior findings showed that tracking longitudinal glucose and ketone trajectories, indicative of glucose-ketone fuel switching, promotes engagement with lifestyle interventions. Assessing dynamic metabolic switching provides deeper insight into the body's adaptive capacity and early indicators of metabolic decline.

[0086] In one exemplary embodiment, metabolic dynamics and long-term outcomes were evaluated in a real-world setting through longitudinal monitoring of a representative subject, designated as N001 (see Figure 5). The subject adhered to a structured health regimen comprising intermittent fasting, structured fitness training, a Mediterranean-style diet, and standardized sleep routines. A comprehensive panel of biomarkers relevant to cardiometabolic health, gut microbiome composition, physical fitness, and sleep quality was continuously or periodically monitored throughout the intervention period. Cardiometabolic parameters, including blood glucose and ketone body concentrations, were obtained from fingerstick and wearable biosensors (e.g., Apple Watch, WHOOP, Garmin). Gut microbiome profiles were assessed using an at-home microbiome test kit (Amili), while fitness performance and sleep parameters were concurrently tracked through gym-based equipment and wearable devices. Dynamic changes are observed in biomarker response when health regimen varies.

[0087] Analysis of longitudinal data from subject N001 revealed characteristic trajectories of glucose and ketone dynamics indicative of glucose-ketone fuel switching, a process reflective of metabolic flexibility. Temporal variations in these biomarkers provided quantitative insights into the subject's adaptive metabolic response under fasting and exercise conditions. Derived parameters such as metabolic switching latency, glycolytic onset rate, and ketotic recovery rate served as individualized indicators of metabolic resilience. Notably, the rate and magnitude of these changes were observed to correlate with the duration and intensity of training, demonstrating that metabolic switching rates may reflect both intervention adherence and physiological adaptation.In further embodiments, digital resilience biomarkers may be derived from a broad range of biological and physiological measurements, including but not limited to glucose, ketone bodies (β-hydroxybutyrate, acetoacetate, acetone), lactate, triglycerides, cholesterol, and inflammatory proteins (e.g., C-reactive protein (CRP)), as well as other cardiometabolic indicators (e.g., homocysteine, apolipoprotein A, apolipoprotein B, and haemoglobin Ale). These biomarkers may be obtained from diverse biological samples such as blood, urine, sweat, saliva, tears, hair, or microbiomes (e.g., skin, oral, nasal, vaginal, urogenital, placental, or gut).

[0088] The system may further employ inferential or predictive algorithms to derive certain biomarkers or biomarker ratios from digital and wearable sensor data (e.g., heart rate variability, sleep stages, or activity levels), thereby reducing reliance on invasive sampling and enhancing real-time dynamic monitoring capabilities. Conversely, biochemical measurements may be used to validate or refine estimates inferred from digital data.

[0089] In a first-in-kind, prospective N-of-1 interventional trial (NUS-IRB-2024-397), various biomarkers were longitudinally collected from subject N001. Following the intervention regimen outlined in Figure 5, the subject's glucose-ketone dynamics were continuously monitored to characterize metabolic adaptation and fuel switching responses. Particular attention was given to biomarker dynamics during strength-focused fitness interventions. During these sessions, blood glucose and ketone levels were periodically measured at approximately 3-15 minute intervals using a fingerstick device, enabling fine-grained temporal profiling of metabolic transitions under exercise-induced physiological stress.

[0090] Referring to both Figures 5 and 6, the biomarkers and biomarker dynamics can be used to calculate a global resilience index test (GRIT) score, a composite resilience score, from which a GRIT age (i.e., biological age as opposed to chronological age) can be calculated. The method 100 involves determining a composite physiological resilience score of a user, by (step 602) receivinglongitudinal data of the user as discussed with reference to step 102. The biomarker data may include, but is not limited to, measurements of glucose, ketone, homocysteine level, apolipoprotein B / apoli poprotein A (ApoB / ApoA) ratio, high-sensitivity C-reactive protein (hsCRP) levels, and homeostatic model assessment of insulin resistance (HOMA-IR) levels. Step 604 then involves generating a resilience score for each biomarker or groups of biomarkers. Step 604 may involve using a piecewise scoring function defined for the corresponding biomarkers.

[0091] For each biomarker, the system computes a respective resilience score using a piecewise scoring function defined for that biomarker or biomarker set. Each piecewise function assigns a resilience score within a defined range of measured values, thereby capturing both physiological variability and clinical significance. Per step 606, the individual resilience scores are then aggregated to compute a GRIT score, which represents an overall measure of the user's physiological resilience or adaptive capacity. The GRIT score may subsequently be used to infer additional parameters, such as a resilience-based biological age, or to provide feedback for adaptive health interventions.

[0092] In some embodiments, one or more tunable parameters of the piecewise scoring functions used for computing biomarker resilience scores may be calibrated via a rank-based optimization procedure that maximizes concordance between computed resilience scores and expert-provided rankings across a calibration dataset of fasting-refeeding cycles. The calibrated parameters may then be fixed and applied to subsequent users or datasets to ensure scoring consistency and generalizability across populations.

[0093] In certain embodiments, the calibrated parameters may comprise a reference spike magnitude and a maximum resilience bonus, which respectively define bounds for a clinically meaningful post-fast rise in biomarker levels and for the contribution of refeeding recovery to the overall resilience score. These parameters may be determined during the calibration phase and subsequentlyfixed for application to other subjects or datasets, ensuring consistent and interpretable resilience scoring across populations.

[0094] Figure 7 illustrates the glucose-ketone dynamics of subject N001 during a strength-centric workout performed on the second day of a 48-hour fast. The shaded area denotes the workout duration. Symbol 1 marks the start of exercise, symbol 2 indicates the glucose-ketone index (GKI) peak, and symbol 3 denotes the recovery point where ketone levels cease increasing postexercise. Abbreviations: Glu - glucose; Ket - ketone; GKI - glucose-ketone index.

[0095] Across both sessions, consistent metabolic patterns were observed. Upon exercise initiation, glucose levels increased and ketone levels decreased, reflecting activation of glycogenolysis and a rapid metabolic shift away from ketosis in response to physical stress. Following exercise cessation, ketone levels gradually rebounded for approximately 100 minutes before stabilizing, indicating sustained post-exercise ketotic recovery. Throughout the 48-hour fasting period and workout sessions, ketone levels remained within a physiologically safe range (<3 mmol / L).

[0096] Substrate switching dynamics were quantified by calculating rates of change in glucose and ketone concentrations across three defined phases: (i) glycolytic onset during exercise, (ii) post-exercise ketotic recovery, and (iii) the complete exercise-recovery period. The Euclidean norm of glucose rate (RGIU) and ketone rate (R. Ket) was applied to capture the combined magnitude of substrate switching, as summarized in Table 1.

[0097] Ketotic

[0098] W Time Time to Glycolytic Onset

[0099] or Recovery Rate Total Substratekout to GKI ketone Rate

[0100] Session duratio (GKI peak -» Switch Rate (n (Exercise start peak recovery Ketone Entire sesion) (min) GKI peak) (min) (min) recovery) (mmol / L / hr) (mmol / L / hr) (mmol / L / hr)

[0101] 2.31 0.75 1.17 (RGIU=2.21, 1 (20241010) 80 57 153 (RGIU=1.09, (RGIU=- 0.67,

[0102] RKet=— 0.68) RKet=0.35) RKet=0.44)

[0103]

[0104] 0.86 0.35 0.53

[0105] 2 (20241025) 45 62 103 (RGIU=0.46, (RGIU=— 0.06, (RGIU=0.21,

[0106]

[0107] RKet=— 0.73) RKet=0.35) RKet=0.49) Table 1. Analysis of metabolic switches based on changing rates of glucose and ketone.

[0108] In this study, the intensity of exercise appeared to influence the glucose-ketone trajectories. In Session 1, a longer workout duration was accompanied by more pronounced substrate-switching rates, whereas Session 2, characterized by a shorter exercise period, exhibited attenuated switching dynamics. These findings suggest that the magnitude of metabolic switching correlates with training load, whereby greater physical demand enhances substrate turnover. Notably, the rate of ketone decline during exercise consistently exceeded the rate of post-exercise ketone recovery, indicating an asymmetry in substrateswitching kinetics, marked by a rapid exit from ketosis in response to acute physical stress, followed by a more gradual re-entry into ketosis during recovery.

[0109] Homocysteine, a sulfur-containing amino acid formed during methionine metabolism, serves as a biomarker of impaired methylation and systemic oxidative stress, both of which contribute to biological aging. Elevated homocysteine levels have been associated with increased cardiovascular risk, cognitive decline, and age-related disorders. During the fast-refeed cycle of a single individual undergoing a structured challenge-recovery protocol, plasma homocysteine levels rose sharply following a 48-hour fast and subsequently declined to near-baseline after approximately 2.5 days of refeeding, shown in Figure 8. Vitamin B complex supplementation was discontinued during the 48h fasting period and resumed following the completion of the fast. This transient elevation may reflect increased protein catabolism and reduced methylation capacity during prolonged fasting, with subsequent normalization upon nutrient reintroduction. The observed pattern underscores the dynamic and reversible nature of metabolic stress responses involving homocysteine.

[0110] Quantification of metabolic switching rates from glucose-ketone dynamicsIn some embodiments of the methods 100 and 600, multiple phases (e.g., phases of activity corresponding to glycolytic onset and ketone recovery) of resilience biomarker behaviour were computed to characterise substrate dynamics. In this embodiment, three phase-specific metabolic switching rates were computed to characterize substrate dynamics in response to exercise: The Glycolytic Onset Rate was defined as the combined rate of glucose increase and ketone decline from the start of exercise to the time of GKI peak. GKI was calculated as follows:

[0111] mmol

[0112] Glucose

[0113] GKI = -,

[0114] (mmolx

[0115]

[0116] Ketone ( — — J

[0117] Glycolytic Onset Rate was calculated as follows:

[0118] A A KGIU=- - - 77“ 7 RKet = —r- - 7-7

[0119] A A

[0120] Glycolytic Onset Rate = J (RGiu2+ RKet2)

[0121]

[0122] Where:

[0123] Gluonsetis the change in glucose level from exercise start to GKI peak. Ketonsetis the change in ketone level from exercise start to GKI peak.

[0124] A tonset is the elapsed time from exercise start to GKI peak.

[0125] Ketotic Recovery Rate was calculated as follows:

[0126] .mmol..mmoL

[0127] Zi j A I £ '

[0128] RGIU

[0129]

[0130] (T1**) A

[0131] Ketotic Recovery Rate = (RGlu2+ RKet2)

[0132]

[0133] Where:

[0134] A Glurecoveryis the change in glucose level from GKI peak to post-exercise ketone recovery.

[0135] Ketrecoveryis the change in ketone level from GKI peak to post-exercise ketone recovery.

[0136] trecoveryis the elapsed time from GKI peak to post-exercise ketone recovery.Ketone recovery is defined as the moment when ketone level stops increasing post-exercise.

[0137] The Total Substrate Switch Rate was calculated as follows:

[0138] =|A ^ |AGiurccwery(^)| |A getonsct(^)|+|A Ketrccovcry(^)|

[0139] A A Total Substrate Switch Rate = J (RGlu2+ RKet2)

[0140]

[0141] Where:

[0142] &.ttotalis the total elapsed time from exercise start to post-exercise ketone recovery.

[0143] Computing Global Resilience Index Test (Grit) Score and Grit Biological Age

[0144]

[0145] 1. Metabolic Flexibility (Ketone-Glucose) Score

[0146] The metabolic flexiblity score is a 0-100 composite that integrates speed to reach ketosis under fasting (time to ketosis score) with flexibility to return to ketosis after exercise (time to ketone recovery score).

[0147] a. Time to ketosis score

[0148] Time to ketosis is obtained from the One-Meal-A-Day (OMAD) ketosis entry regimen. For the subject DELTA001, the regimen was conducted over two consecutive days for 6 sessions (Figure 10). On Day 1, the first measurement was taken immediately after consuming the day's single meal, followed by 3-6 additional measurements until bedtime. On Day 2, the subject resumed glucose and ketone monitoring while performing a structured workout session lasting approximately two hours, consisting of yoga, weightlifting, and intermittent sessions of walking and running. This was followed by a recovery period lasting 2.5 to 5 hours to ensure ketosis state is entered. Time to ketosis is defined as the duration from post-meal measurement on day 1 to post-workout ketosis entry (blood ketone level reaches > 0.5 mmol / L) on day 2.

[0149] b. Time to ketone recovery scoreTime to ketone recovery is obtained from the 48-hour fasting strength training regimen as described in Table 1 and Figure 7. Time to ketone recovery is defined as the elapsed time from GKI peak to post-exercise ketone recovery. Ketone recovery is defined as the moment when ketone level stops increasing post-exercise.

[0150] Computation method of Metabolic flexibility score:

[0151] Time to ketosis score (0-50) = 52.5 - 2.5 * Time to ketosis (Days) Time to ketone recovery score (0-50) = 65 - 10 * Time to ketone recovery (Hours)

[0152] Metabolic flexibility score = Time to ketosis score + Time to ketone recovery score

[0153] DELTA001 Metabolic Flexibility Score:

[0154] ■ Time to Ketosis: 16 hrs 34 mins

[0155]

[0156] 50 pts (based on Figure 9a;

[0157] left)

[0158] ■ Time to Ketone Recovery: 103 mins -> 47.83 pts (based on Figure 9b; right)

[0159] ■ Composite Metabolic Flexibility Score = 50 + 47.83

[0160]

[0161] 97.83 pts

[0162] 2. Methylation Resilience (Homocysteine) Score

[0163] Methylation resilience score is a 0-100 composite score that integrates overall homocysteine level (Base Score) with recovery performance (Resilience Bonus)

[0164] Resilience Score = min(100, Base Score + Resilience Bonus')

[0165] Base score is a piecewise-linear score that converts mean AUC homocysteine to a 0-100 scale by applying graduated reductions across the Elite (<7 pmol / L), Optimal (7-10 pmol / L), and Normal (10-15 pmol / L) bands, with further decreases above 15 pmol / L.

[0166] Base Score (JJ.) — 100,if B < 7Base Score ( ) = 100 — 6(j — 7), if 7 < u < 10 Base Score (g) = 82 — 6.4(g — 10), if g > 10

[0167] Where p is the mean homocysteine level over the fasting-refeeding window, derived from AUC ( I -PreFast + J - PostFasty I *.. „ „,,, (PostFast + PostRFy...

[0168] Number of fasted days + I - - ) * Number of refeeding days

[0169]

[0170] T otal number of fasting and refeeding days

[0171] Figure 11 illustrates the base scoring used to determine the methylation resilience score. In particular, the figure shows a piecewise scoring function that assigns a base score according to the mean area-under-curve (AUC) value of homocysteine levels. The function defines discrete scoring intervals, where different AUC ranges correspond to distinct score segments, as represented by the piecewise equations shown above.

[0172] Spike is the fasting-induced rise in homocysteine, constrained to be nonnegative.

[0173] Spike = max PostFast — PreFast, 0), if (PostFast — PreFast) > 0.5, otherwise 0

[0174] Recovery completeness is the fraction of that spike reversed after refeeding— how fully levels returned toward baseline, clipped to 0-1 (or 0 if Spike=0).

[0175] PostFast — PostRF

[0176] Recovery Completeness = min(max ( - - -, 0), 1),

[0177] Spike

[0178] if Spike > 0; 0 otherwise

[0179] Spike weight is a saturating scale factor that upweights recovery credit for larger fasting spikes while capping their influence.

[0180] (Spike \

[0181] Spike weight = min I — - —, 1 1, So= 3 (imol / L

[0182]

[0183] \ So '

[0184] Sois a reference spike magnitude, or a saturation threshold. Smaller spikes earn proportionally less bonus because there's less to recover from (and more chance it's just noise), while spikes at or above the reference spike magnitude get fullcredit for demonstrating true recovery capability. We chose the value 3 pmol / L because it matches the 3-pmol / L span between the Elite (<7) and Optimal (<10) thresholds, treating a ~3-pmol / L rise as a clinically meaningful "moderate" spike that merits full weighting.

[0185] Resilience bonus is an additive 0-10 point credit that rewards how completely levels return toward baseline after fasting, scaled by the spike's magnitude. We set the max bonus to 10 points— allowing strong recovery to lift the score without overpowering penalties from higher mean values.

[0186] Resilience bonus = Bmaxx Recovery Completeness x Spike Weight

[0187]

[0188] Bma.x — 10

[0189] Computation of subject DELTAOOl's Methylation resilience score

[0190] • PreFast Homocysteine (pmol / L):6.6

[0191] • PostFast Homocysteine (pmol / L): 13.2

[0192] • PostRF Homocysteine (pmol / L): 7.3

[0193] • Mean Homocysteine (from AUC): 10.9

[0194] • Base Score (0-100): 76.2

[0195] • Spike (pmol / L): 6.6

[0196] • Recovery Completeness (0-1): 0.89

[0197] • Reference spike magnitude: 3.00

[0198] • Spike Weight: 1.00

[0199] • Maximum resilience bonus: 10.00

[0200] • Resilience Bonus (0-10): 8.94

[0201] • Final methylation resilience score (0-100): 85.1

[0202] 3. Lipid Regulation Efficiency (ApoA / ApoB) Score

[0203] The total Lipid Regulation Efficiency Score is 100, breaking into two categories:

[0204] a. 20 pts from dynamic changes of ApoB resulting from stressors or interventions (e.g., intermittent fasting).In this case, the ratio of unfasted (3MAD) and fasted (post-48) ratio of ApoB was determined.

[0205] Ratio < 1 = 20 pts

[0206] Ratio > 1 = 0 pts

[0207] DELTAOOl's ApoB Data

[0208] Unfasted Ratio: 74

[0209] Fasted Ratio: 84

[0210] Ratio: 0.88

[0211]

[0212] 20 pts

[0213] b. 80 pts from ApoB / ApoA ratio based on clinical guidelines for men and women. For example, the ideal ratio for men and women is < 0.9 and < 0.8, respectively. Figures 12 illustrates the base scoring scheme for Lipid Regulation Efficiency, in which a piecewise function (as defined by the equations above) provides a score based on the measured ApoB / ApoA ratio.

[0214] Women:

[0215] If ApoB / ApoA ratio < 0.8, Band exposure score = 80

[0216] If ApoB / ApoA ratio > 0.8, Band exposure score = 80 - 5 * (ApoB / ApoA ratio - 0.8)

[0217] Men:

[0218] If ApoB / ApoA ratio < 0.9, Band exposure score = 80

[0219] If ApoB / ApoA ratio > 0.9, Band exposure score = 80 - 5 * (ApoB / ApoA ratio - 0.9)

[0220] DELTAOOl's ApoB / ApoA Ratio Data

[0221] ApoB / ApoA ratio = 0.48 80 pts

[0222] Total Lipid Regulation Efficiency Score = 1004. Inflammation (hsCRP) Score

[0223] The normal range of hsCRP is 1.0-3.0 mg / L, and the optimal (low-risk) range is below 1.0 mg / L. Prior studies have reported that hsCRP levels frequently rose, often significantly, during fasting periods, particularly in obese individuals, and the amount of time takes to restore to baseline varies among individuals. Here, a durable low level of hsCRP during a fasting-refeeding cycle is observed, reflecting strong resilience to 48 hours of fasting. Figure 8 illustrates the inflammation scoring scheme, in which a piecewise function (as defined by the equations above) assigns a score based on the measured high-sensitivity C-reactive protein (hsCRP) level. The function maps hsCRP concentration ranges to corresponding score segments, thereby providing an inflammation score that reflects the degree of systemic inflammatory activity.

[0224] If hsCRP < 0.5, Inflammation Score = 100

[0225] If 0.5 < hsCRP < 3, Inflammation Score = 100 - 20 * (hsCRP - 0.5) If 3 < hsCRP < 5, Inflammation Score = 50 - 25 * (hsCRP - 3)

[0226] If hsCRP > 5, Inflammation Score = 0

[0227] DELTAOOl's hsCRP and Inflammation score

[0228] hsCRP = 0.67 mg / L Inflammation Score = 99.7

[0229] 5. Insulin Sensitivity Score

[0230] Similar to hsCRP, a consistently low level of Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) is observed during a fasting-refeeding cycle. Figure 14 illustrates the insulin sensitivity scoring scheme, in which a piecewise function (as defined by the equations above) assigns a score based on the measured Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) level. HOMA-IR values between 0.5 and 1.4 are considered normal, >1.9 are indicative of early IR, and >2.9 indicate IR

[0231] If HOMA-IR < 1, Insulin sensitivity score = 100

[0232] If 1 < HOMA-IR < 1.4, Insulin sensitivity score = 100 - 5* (HOMA-IR - 1) If 1.4 < HOMA-IR < 1.9, Insulin sensitivity score = 80 - 6* (HOMA-IR - 1.4)If 1.9 < HOMA-IR < 2.9, Insulin sensitivity score = 50 - 4.545* (HOMA-IR - 1-9) If HOMA-IR > 2.9, Insulin sensitivity score = 0

[0233]

[0234] DELTAOOl's HOMA-IR Data and Insulin sensitivity score HOMA-IR = 0.48 mg / L - Insulin sensitivity Score = 100

[0235] 6. Composite GRIT Score

[0236] The GRIT score is then calculated based on two or more of the resilience biomarkers. In some embodiments, the five resilience biomarkers are aggregated. Aggregation may involve adding the scores together, averaging scores, using a weighted average (e.g., where particular resilience biomarkers are affected by socio-demographic factors or genetic factors, which may affect their relevance).

[0237] Global Resilience Index Test (GRIT) for DELTA001

[0238] Metabolic Flexibility Score 97.83

[0239] Methylation Resilience Score 85.1

[0240] Lipid Regulation Efficiency Score 100

[0241] Inflammation Score 99.7

[0242] Insulin Sensitivity Score 100

[0243] GRIT Composite Score for DELTA001

[0244] Average GRIT Composite Score = 96.6

[0245]

[0246] Table 2. Computation of composite GRIT score for DELTA001

[0247] 7. GRIT Biological Age

[0248] The GRIT score represents a composite index of physiological resilience derived from a plurality of measured resilience biomarkers. In one embodiment, the GRIT score is computed based on biomarker measurements including two or more of glucose, ketone, homocysteine level, apolipoprotein B / apolipoprotein A (ApoB / ApoA) ratio, high-sensitivity C-reactive protein (hsCRP), and homeostatic model assessment of insulin resistance (HOMA-IR). However, other relevant resilience biomarkers may additionally or alternatively be incorporated. Thecomposite GRIT score is obtained by averaging the individual resilience scores derived from each measured biomarker (see Figures 7-14). It should be appreciated that the GRIT score may be computed using fewer than all of the listed biomarkers, depending on data availability.

[0249] After obtaining a GRIT score within a range of 0 to 100, a GRIT biological age may be estimated as a resilience-based indicator of biological aging. For example, a GRIT score near 100 may correspond to a biological age approximately 15 years younger than the subject's chronological age, whereas a GRIT score between 0 and 10 may indicate accelerated biological aging relative to chronological age. A GRIT score of approximately 50 may represent parity between biological and chronological age.

[0250] Based on the GRIT score derived from a range of biomarkers (not limited to the above-mentioned biomarkers), a GRIT biological age can be estimated. This estimation is only valid for participants aged 21 or older. The lowest biological age is 18 regardless of the change in age determined from Figure 15.

[0251] The GRIT Biological Age may be determined based on a piecewise linear function applied to the GRIT score. For example, the GRIT Biological Age may be determined from the equations below:

[0252] If 70 ≤ GRIT score ≤ 100, GRIT biological age = chronological age + (35 – 0.5* GRIT score)

[0253] If 50 ≤ GRIT score ≤ 69, GRIT biological age = chronological age + (17.5 - 0.25 * GRIT score)

[0254] If 0 ≤ GRIT score ≤ 49, GRIT biological age = chronological age + (15 - 0.2 * GRIT score)

[0255] Valid only for age 21 or older; the lowest biological age is 18-year-old.

[0256] Figure 15 illustrates the GRIT Biological Age estimation scheme. After determining an overall GRIT score, the corresponding table 3 provides anestimated deviation in biological age relative to the individual's chronological age. The estimation is derived from the equations shown above, which model the relationship between GRIT score and age adjustment. The calculation is valid only for participants aged 21 years or older, with the minimum assignable biological age capped at 18 years.

[0257] DELTA001 is 45-year-old (chronological age) and has an average GRIT composite score of 96.6. Based on Figure 15 and the GRIT biological age equations above, DELTA001 has a GRIT biological age of 31.7 (summarized below).

[0258] GRIT Biological Age for DELTA001

[0259] Chronological Age 45

[0260] GRIT Composite Score 96.6

[0261] Change in Age -13.3

[0262] GRIT Biological Age 31.7

[0263]

[0264] Table 3. Computation of GRIT biological age for DELTA001

[0265] Notably, the GRIT biological age for DELTA001 aligned with that derived from a population dataset containing anonymized biomarker data (U. S. National Health and Nutrition Examination Survey (NHANES), 2021-2023 cycle). The biological age of the Asian population sample from NHANES was estimated using ordinary least square (OLS) regressions by eliminating insignificant variables from variables for which there is clinical evidence that they affect biological aging. The model includes up to 10 variables listed below, with the respective biological age estimates for each individual based on these 10 variables (Table 3). In the case of DELTA001, the estimated biological age using his health metrics from April 2025 was 31.8 years (Stata 18; StataCorp LLC), close to his GRIT biological age. Note that the health metrics used for OLS regression-based biological age estimation are snapshot biomarkers that indicates risk of cardiovascular disease. Currently there is no population dataset that evaluates longitudinal biomarker responses to stressors or interventions as proposed in this inventiondisclosure. The use of OLS regression-based biological age estimation based on snapshot health metrics serves as a reference for the GRIT biological age proposed.

[0266] DELTA001's health metrics for OLS regression-based biological age estimation HbA1c (%): 4.6

[0267] Weight (kg): 74

[0268] Waist circumference (cm): 73.7

[0269] hsCRP (mg / L): 0.28

[0270] HDL (mg / dL): 65

[0271] Systolic blood pressure (mmHg): 109

[0272] Diastolic blood pressure (mmHg): 63

[0273] Pulse rate (bpm): 57

[0274] Gender (Male = 1, Female = 2): 1

[0275] Chronological age (years): 45

[0276] OLS regression-based biological age (years): 31.8

[0277] Analysing temporal variations and interrelationships among these biomarkers enables the computation of dynamic resilience parameters that reflect the user's adaptability to physiological and behavioural stressors or interventions. Such parameters provide insights into metabolic flexibility and overall system resilience that static biomarker readings cannot reveal. Collectively, the present invention enables the definition, computation, and longitudinal monitoring of functional digital resilience biomarkers to support the development of personalized digital twin models. These models simulate and optimize individual health trajectories, enabling predictive, adaptive, and intervention-guided management of metabolic health and healthy longevity.

[0278] The reference in this specification to any prior publication (or information derived from it), or to any matter which is known, is not, and should not be taken as an acknowledgment or admission or any form of suggestion that that prior publication (or information derived from it) or known matter forms part ofthe common general knowledge in the field of endeavor to which this specification relates.

Claims

Claims1. A computer-implemented method for providing a digital twin for improving physiological resilience of a user, comprising steps of: receiving longitudinally collected user data comprising a health regimen of the user and, from a plurality of sensors, measurements indicative of biomarker levels of the user during performance of the health regimen, stressor, or intervention;computing at least one resilience biomarker based on the user data; and generating the digital twin for the user, the digital twin being configured to generate:a health intervention based on the user data and the at least one resilience biomarker, the health intervention being selected to achieve a predetermined change in the at least one resilience biomarker; and a predicted dynamic change in the at least one resilience biomarker corresponding to adherence to the health intervention.

2. The computer-implemented method of claim 1, further comprising: periodically receiving updated user data comprising adherence to the health intervention and further measurements of the biomarker levels from the plurality of sensors; andupdating the digital twin, thereby to update the health intervention, based on the updated user data.

3. The method according to claim 1 or 2, further comprising generating a visualization for a graphical user interface, by a gamification module, to visualize the digital twin, wherein the visualization comprises real-time dynamic feedback indicative of the predicted change in the at least one resilience biomarker based on adherence to the health intervention by the user.

4. The method of according to claim 1 or 3, wherein the method isimplemented in or integral with a wearable device.

5. The method according to any one of claims 1 to 4, wherein the measurements comprise intermittent or periodic measurements of one or more of all anthropometric markers (e.g., body weight, body mass index, blood pressure, and body fat composition), all cardiometabolic biomarkers (e.g., glucose, ketone, homocysteine, apolipoprotein A, apolipoprotein B, haemoglobin Ale, and blood lipids), all inflammatory biomarkers (e.g., interleukin-6, c-reactive protein), all genetic biomarkers (e.g., methylation patterns, telomere length, and genetic profile), and all microbiome biomarkers (e.g., gut, skin, oral, and vaginal).

6. The method according to any one of claims 1 to 5, wherein the digital twin comprises an Al-powered predictive analytics model, trained using baseline measurements of the biomarker levels for a plurality of users, longitudinal measurements of the biomarker levels during performance of at least one predetermined health regimen, stressor, or intervention, to learn relationships between longitudinal changes in the biomarker levels during performance of the health regimen, stressor, or intervention, and output a said predetermined health regimen predicted to achieve an improvement in a resilience biomarker calculated from the longitudinal changes in the biomarker levels, the health intervention corresponding to the said predetermined health regimen.

7. The method according to claim 6, wherein, for the plurality of users, the relationships are relationships between the at least one predetermined health regimen and metabolic state, blood glucose level, ketone profiles, cardiometabolic health and aging biomarkers.

8. The method according to claim 1, wherein the sensors include a wearable sensor and a tissue sampling device.

9. A system for providing a digital twin for improving physiological resilience of a user, comprising:memory;at least one processor;at least two sensors; andthe digital twin,the memory storing instructions that, when executed by the at least one processor, cause the system to:receive longitudinally collected user data comprising a health regimen of the user and, from a plurality of sensors, measurements indicative of biomarker levels of the user during performance of the health regimen;compute at least one resilience biomarker based on the user data; andgenerate the digital twin for the user, the digital twin being configured to generate:a health intervention based on the user data and the at least one resilience biomarker, the health intervention being selected to achieve a predetermined change in the at least one resilience biomarker; anda predicted change in the at least one resilience biomarker corresponding to adherence to the health intervention.

10. The system of claim 9, the at least one processor being configured to:periodically receive updated user data comprising adherence to the health intervention and further measurements of the biomarker levels from the plurality of sensors; andupdate the digital twin, thereby to update the health intervention, based on the updated user data.

11. The system according to claim 9 or 10, further comprising a gamification module and a display, the gamification module being configured togenerate a visualization, for presentation on the display, to visualize the digital twin, wherein the visualization comprises real-time dynamic feedback indicative of the predicted change in the at least one resilience biomarker based on adherence to the health intervention by the user.

12. The method of according to claim 9 or 11, wherein the method is implemented in or integral with a wearable device.

13. The system according to any one of claims 9 to 11, wherein the measurements comprise intermittent or periodic measurements of one or more of all anthropometric markers (e.g., body weight, body mass index, blood pressure, and body fat composition), all cardiometabolic biomarkers (e.g., glucose, ketone, homocysteine, apolipoprotein A, apolipoprotein B, haemoglobin Ale, and blood lipids), all inflammatory biomarkers (e.g., interleukin-6, c-reactive protein), all genetic biomarkers (e.g., methylation patterns, telomere length, and genetic profile), and all microbiome biomarkers (e.g., gut, skin, oral, and vaginal).

14. The system according to any one of claims 9 to 13, wherein the digital twin comprises an Al-powered predictive analytics model, trained using baseline measurements of the biomarker levels for a plurality of users, longitudinal measurements of the biomarker levels during performance of at least one predetermined health regimen, to learn relationships between longitudinal changes in the biomarker levels during performance of the health regimen, and output a said predetermined health regimen predicted to achieve an improvement in a resilience biomarker calculated from the longitudinal changes in the biomarker levels, the health intervention corresponding to the said predetermined health regimen.

15. The system according to claim 14, wherein, for the plurality of users, the relationships are relationships between the at least one predetermined health regimen and metabolic state, blood glucose level, ketone profiles,cardiometabolic health and aging biomarkers.

16. The system according to claim 9, wherein the sensors include a wearable sensor and a tissue sampling device.

17. The system according to any of claims 9, 13, and 16, wherein the biomarker data used for computation of the digital twin comprises both directly measured biomarker values and derived or inferred biomarker values, the derived or inferred biomarker values being generated through algorithmic or computational modelling based on one or more correlated biomarkers.

18. A computer-implemented method for determining a global resilience index test (GRIT) composite physiological resilience score, comprising: receiving, from one or more sensors or data sources, biomarker data comprising longitudinal measurements of a plurality of biomarkers of a user taken during performance of a health regimen, stressor, or intervention;generating for each said biomarker or groups of said biomarkers, a respective resilience score based on a piecewise scoring function defined for the corresponding biomarker; andaggregating the resilience scores to compute the GRIT score the GRIT score representing an overall physiological resilience of the user.

19. The method according to claim 18, further comprising determining a GRIT biological age based on the GRIT score and a chronological age of the user.

20. The method according to claim 18, wherein the biomarker data for computation of the GRIT score comprises measurements of at least one of a glucose, ketone, homocysteine level, apolipoprotein B / apolipoprotein A (ApoB / ApoA) ratio, high-sensitivity C-reactive protein (hsCRP) level, and homeostatic model assessment of insulin resistance (HOMA-IR) level.

21. The method according to claim 1, wherein the digital twin implements the method of claim 18 to determine the GRIT score, the predicted change corresponding to an improved GRIT score.

22. The method according to claim 18, wherein one or more tuneable parameters of the piecewise scoring function for a biomarker are calibrated via a rank-based optimization procedure that maximizes concordance between computed resilience scores and expert-provided rankings across a calibration dataset of fasting-refeeding cycles.

23. The method according to claim 18, wherein the calibrated parameters comprise a reference spike magnitude and a maximum resilience bonus defining respective bounds on a post-fast rise in levels of the biomarkers and refeeding recovery contributions.

Images