A motion prescription timing programming method and system based on a dual digital twin and a five-dimensional biokinetic

Abstract

The application provides a motion prescription timing programming method and system based on a dual digital twin and a five-dimensional life, solves a complex health management system collaborative evolution problem, and comprises a multi-source data management module, a body-mind dual digital twin engine, a prescription optimization generation module, a simulation deduction decision module and a self-adaptive execution and evolution module, and each module communicates through API or a message queue.The application has the advantages of high intelligent level, good safety and the like.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary fields of digital health, precision sports medicine and artificial intelligence, and specifically relates to a method and system for timing programming of exercise prescriptions based on dual digital twins and five-dimensional generative mechanisms. Background Technology

[0002] Personalized exercise prescriptions are a core technological means to improve the effectiveness of health interventions and ensure exercise safety. However, existing exercise prescription generation technologies have fundamental limitations in terms of data fusion depth, decision model dimensions, and system architecture, making it difficult to meet the complex demands of the era of precision health management, which prioritizes safety and achieves overall optimization.

[0003] First, at the data fusion and cognitive level, existing technologies largely rely on macroscopic physiological indicators and subjective questionnaire data, lacking the ability to integrate microscopic molecular mechanisms and cross-dimensional generative and restraining network effects. Cutting-edge scientific evidence suggests that the health effects of exercise are not limited to a single target system, but rather generate cross-system synergistic and antagonistic effects through complex inter-organ dialogue. For example, studies by Song et al. and Newton et al. revealed a positive synergistic effect between exercise (corresponding to wood in the five dimensions) and the immune system (corresponding to metal), i.e., the balance of metal overcoming wood in the exercise context; while Huang et al.'s research found that excessively strenuous exercise can damage hippocampal function through the muscle-brain axis, empirically demonstrating the risk of excessive exercise (wood) inhibiting the neurocognitive system (earth) through the wood-overcoming-earth principle. These key findings revealing the bidirectional interaction effects between the five dimensions have not yet been translated into a computable and integrable quantitative model of the five-dimensional generative and restraining network.

[0004] Second, at the level of model building and decision-making paradigms, existing digital twin or intelligent recommendation systems mostly focus on simulating the positive benefits of a single target system, lacking a safety-priority decision-making framework embedded with a five-dimensional mutual generation and restraint network constraint. Current technologies cannot quantitatively model the aforementioned mutual generation and restraint relationships such as wood restraining earth and fire generating earth supporting each other, let alone embed them as rigid constraints into the core logic of prescription generation algorithms. This leads to prescription generation often aiming to maximize local physiological benefits while neglecting the potential hidden and cumulative damage to other key physiological dimensions, resulting in a significant blind spot in system security.

[0005] Third, in terms of system architecture and evolutionary capability, existing solutions are mostly isolated tools that provide static suggestions. They cannot receive and analyze global security optimization goals from upper-level health management strategies, nor can they continuously verify and optimize their prediction accuracy of the five-dimensional Mutanh effect through long-term, multi-dimensional closed-loop feedback. They also lack self-evolutionary capability.

[0006] In summary, how to construct an intelligent prescription time-series programming system that can deeply integrate multi-source evidence, use an endogenous five-dimensional generative-kinesthetic network dynamics model as the core computing engine, and co-evolve with complex health management systems has become a core challenge in overcoming current technological bottlenecks. Summary of the Invention

[0007] The purpose of this invention is to address the above-mentioned problems by providing a method and system for timing programming of exercise prescriptions based on dual digital twins and five-dimensional genocide, which enables timing programming, safety verification, and dynamic control of exercise prescriptions while adhering to individual biological foundations and global safety objectives.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: a motion prescription timing programming method based on dual digital twins and five-dimensional sense, comprising the following steps: S1: Multi-source data fusion and twin initialization The system acquires multi-source health data from users and performs personalized initialization of pre-constructed system digital twins and cognitive digital twins based on the data; the system digital twin integrates a five-dimensional generative network sub-model for simulating the interaction between different physiological dimensions.

[0009] Multi-source health data includes at least the user's static multi-omics baseline data and dynamic health assessment data. The initialization process deeply calibrates the core physiological parameters and metabolic network status of the system's digital twin, particularly the sensitivity parameters of the five-dimensional genotype network based on the latest scientific discoveries, using pre-defined scientific mapping rules. For example, it sets the potential baseline for the user's wood axis (motor system) based on their ACTN3 genotype; and it jointly calibrates the initial sensitivity parameters of the wood-soil axis (muscle-brain axis) sub-model based on IL-6 polymorphism sites and baseline metabolomics data. The cognitive digital twin is initialized synchronously, mapping the user's training preferences and behavioral motivations. The initialization process can also receive and integrate personalized health baseline parameters and phased priority protection dimension target instructions from the upper-level management system.

[0010] Furthermore, the scientific mapping rules are a rule base constructed based on the association analysis results of large-scale population cohorts and biological mechanism studies. Specifically, for genomic data, a multi-gene risk scoring method is used to combine multiple related loci into a comprehensive sensitivity index; for metabolomics data, key metabolites are identified through metabolic pathway enrichment analysis and mapped to the functional state of the corresponding physiological system; for microbiome data, metagenomic functional prediction is used to calculate the activity of specific metabolic pathways as proxy indicators of immune and inflammatory states. The sensitivity parameters of the five-dimensional Sinker network are specifically represented by a set of adjustable weight coefficients and thresholds. For example, the initial sensitivity parameters of the wood-soil axion model can be expressed as: S_mt = f(IL6_rs1800795, tryptophan level, kynurenine level), where f is a mapping function pre-trained by logistic regression or a neural network.

[0011] S2: Dynamic State Awareness and Policy Reception It acquires macroscopic motion decision commands from the upper-level time-series decision engine and real-time monitoring data of the user, and integrates them with the current status data output by the dual digital twins to obtain the user's multi-dimensional comprehensive status, forming a comprehensive perception of the user's current five-dimensional health status.

[0012] Macroscopic motion decision-making instructions are transmitted in structured data objects such as JSON or Protocol Buffers, containing at least the target dimension (e.g., wood), target intensity coefficient (0.5-1.5), and a risk constraint dictionary (e.g., wood overcoming earth upper limit: 0.6, metal overcoming wood upper limit: 0.5). Real-time monitoring data includes physiological signals such as heart rate, acceleration, and skin conductance collected by wearable devices, as well as psychological self-assessment data (e.g., fatigue level, mood score) obtained through smartphones or intelligent voice assistants. The fusion process employs Kalman filtering or multi-sensor fusion algorithms to align heterogeneous data under a unified timestamp and update the state variables of the two digital twins.

[0013] S3: Multi-objective optimization of prescriptions with integrated five-dimensional Sinker-Kruger network constraints The multidimensional integrated state is input into a multi-objective optimization model to generate one or more candidate exercise prescriptions. The constraint library of the multi-objective optimization model contains five-dimensional Sinker effect quantitative constraint values ​​calculated by the five-dimensional Sinker network sub-model based on the current multidimensional integrated state, which are used to quantitatively evaluate the potential negative impact of exercise on other non-targeted physiological dimensions.

[0014] The optimization objective function should include at least the physiological benefit score predicted by the system's digital twin and the psychological acceptability score predicted by the cognitive digital twin. The constraint library system integrates multiple types of constraints, the most crucial of which are the five-dimensional generative and restraining effect quantitative constraints, such as the Wood-Earth neurocognitive inhibition risk value and the Metal-Wood immune perturbation risk value, calculated in real-time by the five-dimensional generative and restraining network sub-model. When solving the optimization model, it is essential to ensure that all candidate solutions satisfy the personalized dynamic equilibrium domain constituted by the aforementioned network constraint values.

[0015] The multi-objective optimization model employs constrained evolutionary algorithms such as NSGA-II, MOEA / D, or Bayesian optimization for solution. Optimization variables include exercise type, intensity, duration, frequency, and interval ratio. After each candidate solution is generated, the risk value is calculated using the five-dimensional Sinker network sub-model in the system's digital twin: for the Mu-Tu risk, the sub-model can be expressed as a set of differential equations: dC_brain / dt = k1 * F_muscle(t) - k2 * C_brain, where F_muscle(t) is the muscle factor release function, k1 is related to the user's neuroinflammatory sensitivity parameter; the risk value is defined as the integral of C_brain exceeding the individual threshold. The constraint library also includes real-time recovery state constraints based on dynamic metabolomics data, such as blood lactate clearance rate, and inflammatory load constraints based on the microbiome, such as the lipopolysaccharide potential index. These constraints are stored in the user's digital profile in the form of personalized dynamic thresholds.

[0016] S4: Cross-level digital twin collaborative simulation and security decision-making One or more candidate exercise prescriptions are input into an initialized dual digital twin for collaborative simulation. The simulation of the system's digital twin is mandated to include an assessment of the activation level of the five-dimensional generative-relationship network effects (such as the wood-earth axis and the metal-wood axis) and their impact on distal dimension functions. The cognitive digital twin simultaneously simulates the user's psychological acceptance. Based on a pre-defined comprehensive evaluation function, a final exercise prescription is selected from the candidate prescriptions. This function explicitly incorporates the five-dimensional generative-relationship network balance score as a core negative weight to prioritize avoiding network imbalance solutions. Finally, the prescription with the optimal overall expected return and controllable dynamic balance of the five-dimensional generative-relationship network is selected as the final exercise prescription, and a corresponding personalized guidance strategy script is generated.

[0017] The multi-scale simulation of the system's digital twin employs a hierarchical modeling and coupling strategy: the molecular layer uses reaction-diffusion equations or Boolean networks to simulate the activation dynamics of key pathways such as AMPK and PGC-1α; the cellular layer uses a surrogate-based model to simulate the proliferation and differentiation of muscle satellite cells; the tissue layer uses a finite element model to calculate the biomechanical response of the musculoskeletal system; and the system layer uses a compartmentalized model to describe the distribution and clearance of hormones and cytokines. The five-dimensional biomechanical network effect assessment module takes the outputs of each layer as input, calculates the state changes and biomechanical relationship indicators of each dimension, and finally outputs a five-dimensional network balance score. This score can be defined as a weighted sum of the risk values ​​of each biomechanical path, or as a measure based on the distance between the multidimensional state space and the ideal equilibrium point. The comprehensive evaluation function is specifically in the form: F = w1·physiological benefit - w2·network balance score + w3·psychological score, where w1, w2, and w3 are adjustable weights. The default values ​​are calibrated using expert knowledge and historical data, and can be personalized and optimized through the evolutionary process in step S5.

[0018] S5: Dual-Loop Adaptive Execution and System Evolution During the user's execution of the final exercise prescription, exercise parameters are dynamically adjusted based on real-time sensing data, forming a dual-loop regulation of physiological-behavioral and psychological-emotional aspects. Furthermore, multi-dimensional feedback data, including cross-dimensional functional indicators, is used after user execution to iteratively optimize the model parameters of the system's digital twin, particularly calibrating the predictive accuracy of the five-dimensional Sinker network sub-model. This enables the entire decision-making system to continuously evolve in terms of both safety and accuracy.

[0019] The physiological-behavioral regulation loop employs a model-based predictive control strategy: based on real-time heart rate, power output, and other data, it predicts the physiological state for the next few minutes. If the deviation from the expected trajectory, such as an excessively high heart rate, is detected, the intensity target of subsequent motor units is automatically fine-tuned or the recovery time is extended. The adjustment range is constrained by a pre-generated safety rule base. The psychological-emotional regulation loop dynamically selects guiding dialogue templates based on the real-time emotional states output by the cognitive digital twin, such as boredom and fatigue, and optimizes the dialogue selection strategy through reinforcement learning. Multi-dimensional feedback data includes: post-execution cognitive function test results, such as Stroop task reaction time, blood inflammatory markers such as hs-CRP, IL-6, and sleep quality index. These data are compared with the corresponding predicted values ​​of the twin in step S4, the prediction error is calculated, and the parameters of the five-dimensional Schenck network sub-model, such as k1 and k2, are fine-tuned using Bayesian updates or online gradient descent methods to gradually approximate the user's actual response characteristics.

[0020] In the above-mentioned exercise prescription timing programming method based on dual digital twins and five-dimensional synergy, the quantitative constraint value of the five-dimensional synergy effect is the risk value of wood-earth neural cognitive inhibition. It is calculated by the wood-earth axis (muscle-brain axis) signal transduction sub-model in the system digital twin and is used to quantitatively assess the potential negative impact of candidate exercise prescriptions on the function of the central nervous system.

[0021] In the above method, the five-dimensional Sinker effect quantitative constraint value is the risk value of the perturbation of the Jin-Mu immune homeostasis. It is calculated by the Jin-Mu axis (motor-immune axis) signal transduction sub-model in the digital twin of the system and is used to quantitatively assess the potential negative impact of candidate exercise prescriptions on the homeostasis of the body's immune system.

[0022] In the above method, the multi-source health data in step S1 includes the user's genomic data; personalized initialization further includes: setting the initial sensitivity parameters of the corresponding five-dimensional genotyping network sub-model in the system's digital twin based on the gene locus information in the genomic data that is related to the sensitivity of a specific five-dimensional genotyping pathway.

[0023] In the above method, the constraint library of the multi-objective optimization model in step S3 also integrates priority protection target constraints from the upper management system. This instruction is transformed into a lower threshold of the corresponding risk constraint in the system's digital twin, ensuring that prescription generation strictly follows the global safety optimization objective.

[0024] In the above method, the collaborative simulation and deduction of the system digital twin in step S4 is a multi-scale, integrated prediction and simulation from molecular pathway activation, cellular stress response, tissue biomechanical changes to five-dimensional sense network signal transduction and its impact on the function of distal organ systems.

[0025] In the above method, the iterative optimization in step S5 specifically involves comparing the actual measured feedback data with the predicted values ​​of the system digital twin in step S4, and calibrating the internal parameters of the relevant five-dimensional Sinker network sub-model based on the comparison deviation.

[0026] A timing programming system for exercise prescriptions based on dual digital twins and five-dimensional senses includes: The multi-source data management module is used to acquire, store, and process users' multi-source health data. This module has a built-in data cleaning pipeline to perform quality control, normalization, and missing value imputation on raw data such as genomics, metabolomics, and microbiome. It uses feature engineering methods such as principal component analysis and autoencoders to extract low-dimensional feature vectors. It also builds a unified data model to associate multimodal data with a unique user identifier, supporting version management and historical traceability. The mind-body dual digital twin engine is used to construct and initialize the system digital twin and the cognitive digital twin. The system digital twin integrates a five-dimensional generative-relationship network sub-model. The system digital twin employs a hybrid modeling approach: the physiological parameter model describes core physiological processes based on differential equations; the five-dimensional generative-relationship network sub-model uses a graph neural network architecture, with the user's current state as node features and generative-relationship relationships as edges, simulating the mutual influence between dimensions through a message passing mechanism, and its output is a risk score for each generative-relationship path. The cognitive digital twin, based on recurrent neural networks and attention mechanisms, learns the user's exercise preferences, fatigue patterns, and emotional response patterns from historical behavioral data. The prescription optimization generation module integrates a constraint library containing five-dimensional Sinker network constraint rules into its built-in multi-objective optimization model. This module maintains an extensible constraint rule library, where each rule includes trigger conditions, constraint expressions, and penalty weights. During optimization, a genetic algorithm with an elitist strategy is employed, with a population size of 100, a crossover probability of 0.8, a mutation probability of 0.1, and 200 generations. In each generation, each individual (candidate prescription) uses a dual-digital twin to calculate the objective value and constraint violation degree, and selects the Pareto front solution set through non-dominated sorting and crowding distance. The simulation and decision-making module drives two digital twins to collaboratively simulate candidate prescriptions and make safety decisions based on a comprehensive evaluation function that includes a five-dimensional Sinker network to balance negative scores. This module employs a parallel computing architecture, distributing candidate prescriptions to multiple computing nodes for simultaneous simulation, significantly reducing decision latency. The weight coefficients in the comprehensive evaluation function can be learned offline from historical successful or failed cases, or adjusted online based on user feedback. The adaptive execution and evolution module is used to adjust motion parameters based on real-time data during execution and collect feedback data after execution to optimize the system model, especially the five-dimensional Schenck network sub-model. This module includes a real-time event processing engine that subscribes to data streams from wearable devices and uses complex event processing techniques to identify abnormal patterns such as sudden increases in heart rate and cadence irregularities, triggering preset control rules. Feedback data is stored in a time-series database, periodically triggering model retraining tasks. Incremental learning algorithms such as stochastic gradient descent are used to update neural network parameters, while simultaneously recording model versions to support rollback and A / B testing.

[0027] In the above system, the system digital twin is a multi-scale computable model that integrates biomechanical models, metabolic network models, endocrine and immune models, and at least one five-dimensional genocidal network effect model such as the wood-soil axis model or the metal-wood axis model.

[0028] The system also includes a data interface module, which is used to receive quantitative risk data reports from external health assessment services, or to feed back prescription execution results and twin status updates to the upper-level time-series decision engine, so as to realize collaborative operation in complex health management systems.

[0029] The data interface module supports both RESTful API and message queue communication methods, employing OAuth2.0 and JWT for authentication and authorization. Data exchange is standardized to the FHIR (Rapid Healthcare Interoperability Resource) standard, ensuring seamless integration with various health service platforms. Received external reports, such as AI aging clock assessment results, are parsed and sent to the multi-source data management module as part of the initialization data. Data fed back to the upper-layer engine includes prescription execution completion rate, five-dimensional state change summary, and twin confidence scores, used for dynamic adjustment of upper-layer strategies.

[0030] The system also includes a personalized interaction and guidance interface module, which is connected to the adaptive execution and evolution module. This module is used to present users with executable exercise prescriptions packaged with guidance strategies, provide real-time visual guidance and voice / text prompts during execution, and collect users' subjective feedback data.

[0031] Compared with existing technologies, the advantages of this invention are: 1. By combining the concept of mutual generation and restraint in complex systems theory with empirical findings in cutting-edge molecular biology, a quantitative modeling of cross-system antagonistic effects such as the risk of inhibition of neurocognitive function and the risk of immune homeostasis disturbance is proposed. This model is then integrated as a rigid constraint into a multi-objective optimization algorithm, which solves the problem that existing methods only pursue local benefits and ignore the risks of interaction between systems. This elevates the generation of exercise prescriptions from empirical safety to a new level of quantitative prevention. 2. Through the mind-body dual digital twin engine and cross-system interaction sub-model, the biomarker-effect relationship discovered in the latest breakthrough research can be quickly transformed into dynamic parameters or optimization sub-objectives of the digital twin model; through the execution-multi-dimensional feedback closed loop, the system can use real cross-system functional data to continuously verify and optimize its risk prediction model, forming a measurable, verifiable and interpretable technical advantage. 3. Dynamic evolution capability is achieved through dual-loop adaptive execution and system evolution mechanism. During execution, motion parameters are dynamically fine-tuned based on real-time sensor data, and guidance dialogue is adaptively adjusted. At the same time, periodic cross-system function retest data is used as feedback to feed back and iteratively optimize the internal response parameters of the digital twin and the prediction accuracy of cross-system interaction sub-models, so that the system's safety warning capability and personalized accuracy continue to improve with the increase of usage time. 4. Positioned as a secure execution terminal at the system architecture level, it achieves bidirectional communication with the upper-layer time-series decision engine through the data interface module. It can accurately receive and parse global security optimization goals such as prioritizing the protection of neurocognitive functions in the current stage, and can also feed back prescription execution results and twin status updates to the upper layer, thus realizing precise collaborative operation in a complex management framework. Attached Figure Description

[0032] Figure 1 This is a system positioning diagram of the present invention; Figure 2 This is an overall flowchart of the present invention; Figure 3 This is the data fusion and initialization diagram of the present invention; Figure 4 This is the prescription generation diagram of the present invention; Figure 5 This is the simulation and decision-making diagram of the present invention; Figure 6 This is the execution evolution diagram of the present invention; Figure 7 This is the system architecture diagram of the present invention. Detailed Implementation

[0033] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. Example 1

[0034] like Figure 1-7 As shown in the figure, this embodiment initializes the five-dimensional genotyping network sub-model based on genomic data. It mainly demonstrates how to use the user's genomic data to perform personalized initialization of the five-dimensional genotyping network sub-model in the system's digital twin. The specific implementation process is as follows: S1: Data collection. Users provide saliva samples through compliant channels for whole-genome sequencing. After quality control, the sequencing data is input into the multi-omics data management and feature engineering module.

[0035] Furthermore, the scientific mapping rules are a rule base constructed based on association analysis results from large-scale population cohorts and biological mechanism studies. For genomic data, a multi-gene risk scoring method is used to combine multiple relevant loci into a comprehensive sensitivity index; for baseline metabolomics data, key metabolites are identified through metabolic pathway enrichment analysis and mapped to the functional state of the corresponding physiological systems. The five-dimensional Sinker network sensitivity parameters are specifically represented by a set of adjustable weight coefficients and thresholds.

[0036] S2: Gene locus analysis. Based on consensus in sports genetics, the feature engineering module analyzes gene loci related to athletic ability, injury risk, and sensitivity to the five-dimensional serocontrol pathway. Exercise capacity related (corresponding to wood element): Detect the ACTN3 gene rs1815739 site. If it is the R allele (RX or RR type), it is determined that the baseline proportion of type II muscle fibers is higher and the strength training gain coefficient is larger. Set the potential baseline of its wood element (muscle system) to the higher range. If it is type II in the ACE gene I / D polymorphism, it is determined that the potential for aerobic endurance development is higher.

[0037] Damage risk related (corresponding to soil and ligament bearing capacity): Detecting the COL1A1 gene rs1800012 site, if it carries a specific variant related to weak connective tissue strength, then the maximum load threshold of tendons / ligaments needs to be significantly reduced, that is, the baseline structural support capacity of soil and ligaments is reduced.

[0038] Neuroinflammatory sensitivity (corresponding to wood-soil axis sensitivity): Detecting neuroinflammatory-related loci such as IL-6 gene rs1800795, if the genotype is carried and associated with high sensitivity to inflammatory response, the initial sensitivity parameter of the wood-soil axis (muscle-brain axis) sub-model needs to be set to a higher range.

[0039] S3: Initialization of the five-dimensional generative network sub-model; the mind-body dual digital twin engine receives the above analysis results and performs deep initialization and calibration on the system's digital twin: Set the baseline proportion of type II muscle fibers and the strength training gain coefficient to the corresponding range to complete the basic parameter configuration of the muscle fibers; Lower the maximum load threshold of tendons / ligaments to a safe range to complete the setting of soil bearing capacity threshold; The initial sensitivity parameters of the wood-soil axis (muscle-brain axis) sub-model were increased and calibrated together with the levels of tryptophan and neurotransmitter precursors in the baseline metabolome to construct a predictive basis for the central effects of circulatory factors induced by the user's exercise (wood-soil effect).

[0040] Furthermore, the initial sensitivity parameter of the wood-soil axle model can be expressed as: S_mt = f(IL6_rs1800795, tryptophan level, kynurenine level), where f is a mapping function pre-trained by logistic regression or a neural network, used to quantify the user's innate sensitivity to the risk of motion-induced neurocognitive inhibition.

[0041] S4: Upper-level target fusion. The initialization engine simultaneously receives instructions from the upper-level management system {priority protection dimension: earth (neuro-cognition), safety redundancy coefficient: 0.8}. These instructions are transformed into a lower threshold for the corresponding wood-earth axis risk constraint in the system's digital twin, serving as input for subsequent optimization.

[0042] Through the above initialization, the system's digital twin already possesses physiological potential, safety threshold, and five-dimensional Sinker network sensitivity parameters that reflect the user's unique genetic background before generating a prescription, laying the foundation for accurate risk quantification in subsequent steps. Example 2

[0043] This embodiment mainly demonstrates how to generate the Muketu neurocognitive inhibition risk value and integrate it as a constraint into a multi-objective optimization model. The specific implementation process is as follows: S1: Scenario setting: The upper-level temporal decision engine sends macroscopic motion decision instructions to the system, requiring an increase in metabolic capacity (corresponding to the wood system in the five-dimensional model). The current user's multidimensional comprehensive status shows: the user's physiological state is good, but genetic and baseline data indicate a neuroinflammatory sensitive type (i.e., a wood-soil axis highly sensitive type).

[0044] Furthermore, macroscopic motion decision-making instructions are transmitted in structured data objects such as JSON format, containing at least the target dimension (wood), target intensity coefficient, and risk constraint dictionary (e.g., wood over earth upper limit: 0.5). The multidimensional integrated state is obtained by fusing real-time monitoring data with the current state data output by dual digital twins through Kalman filtering or multi-sensor fusion algorithms.

[0045] S2: Candidate prescription generation. After receiving the instruction, the prescription optimization generation module calls the multi-objective optimization model to start solving the problem. The optimization objectives include: Maximize the physiological benefit score (such as energy consumption, mitochondrial biogenesis) predicted by the system's digital twin. Maximize the psychological acceptance scores (such as pleasure and willingness to persist) predicted by the cognitive digital twin.

[0046] Furthermore, the multi-objective optimization model is solved using a constrained evolutionary algorithm such as NSGA-II. Optimization variables include motion type, intensity, duration, frequency, and interval ratio.

[0047] S3: Five-dimensional Mundi network constraint calculation. For each candidate prescription generated during the optimization process, the system's digital twin calls its internal five-dimensional Mundi network sub-model for risk quantification assessment. Taking high-intensity intermittent training candidate prescriptions as an example: Wood-soil axis model call: The model simulates and predicts the release dynamics of muscle factor X induced by exercise based on prescription parameters (intensity, duration, interval ratio); Combining user sensitivity parameters: Based on the high sensitivity parameters initialized in Example 1, the model calculates the risk value of Mukeshvara neurocognitive inhibition that this prescription may cause temporary inhibition of hippocampal mitochondrial function, and the output result is 0.7; Comparison with threshold: The system's preset risk threshold for this user's prescription is 0.5. Since 0.7 is greater than 0.5, the prescription is determined to be outside the safety range.

[0048] Furthermore, the Muketu risk sub-model can be expressed as a set of differential equations: dC_brain / dt = k1 * F_muscle(t) - k2 * C_brain, where F_muscle(t) is the muscle factor release function, k1 is related to the user's neuroinflammatory sensitivity parameter, and the risk value is defined as the integral of C_brain exceeding the individual threshold.

[0049] S4: Constraint Integration. The constraint library for the multi-objective optimization model includes the following rule: the Mukeshvara neural cognitive inhibition risk value corresponding to all candidate solutions must be less than or equal to 0.5. Since the high-intensity intermittent training prescription violates this five-dimensional Mukeshvara network constraint, the optimization algorithm automatically excludes it and continues to search for other feasible solutions.

[0050] Furthermore, the constraint library also includes real-time recovery state constraints based on dynamic metabolomics data and inflammatory load constraints based on the microbiome. These constraints are all stored in the user's digital profile in the form of personalized dynamic thresholds.

[0051] This embodiment embeds the risk value of Muketu neurocognitive inhibition as a rigid constraint into the optimization model. The system ensures that only prescriptions that meet the balance requirements of the five-dimensional Muketu network can enter the subsequent deduction stage, realizing a safety-first decision-making logic. Example 3

[0052] This embodiment demonstrates the cross-level collaborative simulation and deduction of candidate prescriptions and the final decision based on a comprehensive evaluation function that includes a balanced negative score from a five-dimensional Sinker network. The specific implementation process is as follows: S1: Input candidate prescriptions, generating two candidate prescriptions that satisfy all constraints: Prescription A: Moderate-intensity continuous aerobic exercise (60% HRmax, 30 minutes); Prescription B: Low-intensity interval training (alternating between 50% and 70% HRmax, 25 minutes).

[0053] S2: System digital twin simulation and deduction, with two prescriptions input into the system digital twin for multi-scale, integrated prediction simulation: Molecular pathway layer: Simulates the activation levels of pathways such as AMPK and PGC-1α; Cellular response layer: mimicking mitochondrial biogenesis and myosatellite cell activation; Tissue biomechanical layer: Simulates the mechanical response of muscles, bones, and tendons; Five-dimensional generative and restraining network layer: calling the wood-earth axis (muscle-brain axis) sub-model to predict the effect of circulating factors on the central nervous system (wood restrains earth effect); calling the metal-wood axis (motor-immune axis) sub-model to predict changes in the distribution of immune cell subsets (metal restrains wood effect). System Physiological Layer: Integrates the above simulation results and outputs a comprehensive prediction of the impact on the functions of various physiological systems.

[0054] Furthermore, the multi-scale simulation of the system's digital twin employs a hierarchical modeling and coupling strategy: the molecular layer uses reaction-diffusion equations to simulate the activation dynamics of key pathways; the cellular layer uses a surrogate-based model to simulate the proliferation and differentiation of myosatellite cells; the tissue layer uses a finite element model to calculate biomechanical responses; and the system layer uses a compartmentalized model to describe the distribution and clearance of hormones and cytokines. The five-dimensional interaction network effect assessment module uses the outputs of each layer as inputs to calculate state changes and interaction relationship indices in each dimension.

[0055] S3: Cognitive Digital Twin Simulation and Deduction: Two prescriptions are simultaneously input into the cognitive digital twin to simulate psychological states. Simulate the user's subjective experience when executing prescription A, such as being slightly boring but acceptable; Simulate the emotional fluctuations of users when executing prescription B, such as interesting changes and high levels of pleasure; Output a predicted score for the psychological acceptance of each prescription.

[0056] Furthermore, cognitive digital twins, based on recurrent neural networks and attention mechanisms, learn users' exercise preferences, fatigue patterns, and emotional response patterns from historical behavioral data.

[0057] S4: The safety decision module makes a decision. The simulation and deduction decision module calls the preset comprehensive evaluation function to score each prescription. The function format is as follows: F = α·physiological benefit score - β·five-dimensional genocide network imbalance score + γ·psychological acceptability score; The imbalance score of the five-dimensional generative-relationship network can be defined as the weighted sum of the risk values ​​of each generative-relationship path, or as a measure of the distance between the multidimensional state space and the ideal equilibrium point. α, β, and γ are preset weight coefficients, with default values ​​calibrated through expert knowledge and historical data, and can be personalized and optimized through an evolutionary process.

[0058] The calculation results show that: Prescription A: Moderate physiological benefits, low risk of imbalance in the five-dimensional generative and restraining network (0.2 risk of wood restraining earth, 0.2 risk of metal restraining wood), moderate psychological acceptance, and high overall score; Prescription B: Slightly lower physiological benefits, extremely low risk of imbalance in the five-dimensional generative and restraining network (0.1 risk of wood restraining earth, 0.1 risk of metal restraining wood), high psychological acceptance, and a comprehensive score close to but slightly lower than A.

[0059] Based on the principle of prioritizing the avoidance of high-risk options, the decision-making module confirmed that neither prescription was high-risk and selected prescription A, which had the highest comprehensive score, as the final exercise prescription.

[0060] S5: Personalized guidance strategy generation. Based on the simulation results of the cognitive digital twin, the system generates a matching guidance script for prescription A: "Today we embark on a steady cardiopulmonary journey, focusing on the rhythm of your breathing, which will gently boost your metabolic engine while keeping your mind clear." Furthermore, the guidance strategy script is dynamically selected from a preset template library. The template selection is based on the emotional state output by the cognitive digital twin, and the selection strategy is optimized through reinforcement learning.

[0061] This embodiment uses collaborative simulation with two digital twins to comprehensively evaluate the multidimensional effects of candidate prescriptions before the user actually executes them, especially the quantitative evaluation of the five-dimensional Sinker network effect, and makes a safety decision based on a comprehensive evaluation function that includes a penalty term for imbalance in the five-dimensional Sinker network. Example 4

[0062] This embodiment demonstrates the real-time control mechanism during the user's execution of the final exercise prescription, as well as the iterative optimization of the system model using post-execution feedback data, particularly the calibration of the five-dimensional Sinker network sub-model. The specific implementation process is as follows: S1: The execution process is controlled by a dual-loop system. The user executes the prescription A (moderate-intensity continuous aerobic exercise) selected in Example 3, and the intelligent devices (heart rate belt, power meter) monitor physiological data in real time.

[0063] Physiological-behavioral regulation loop: Real-time data stream shows that the user's heart rate is rising too fast. The multimodal perception and adaptive execution closed-loop module determines that the user is not warming up enough. It immediately prompts through a personalized interactive interface: "Please slow down a little and let us find the rhythm first." It also automatically adjusts and reduces the intensity target of the first exercise phase by 5%.

[0064] Psychological-emotional regulation loop: Real-time monitoring of user cadence fluctuations, combined with cognitive digital twin predictions, determines that users may show early signs of boredom, and the interface simultaneously displays encouraging prompts: "Keep this pace, you've completed 1 / 4, how do you feel?" Furthermore, the physiological-behavioral regulation loop employs a model-based predictive control strategy: it predicts the physiological state over the next few minutes based on real-time data, and if the state deviates from the expected trajectory, it automatically fine-tunes the intensity target of subsequent motor units or extends the recovery time. The psychological-emotional regulation loop dynamically selects guiding dialogue templates based on the emotional state output in real time by the cognitive digital twin.

[0065] S2: Post-execution feedback data collection.

[0066] 24 hours after the exercise, users complete a simple cognitive attention test, such as a reaction time test, through the accompanying app, and the results are automatically uploaded to the system.

[0067] The system simultaneously collects user-authorized blood samples and detects changes in specific inflammatory markers such as IL-6 and TNF-α.

[0068] Furthermore, the multi-dimensional feedback data includes: post-execution cognitive function test results, blood inflammatory markers, sleep quality index, etc. These data serve as cross-dimensional functional indicators to verify the predictive accuracy of the five-dimensional Schenck network sub-model.

[0069] S3: Comparison of predicted and measured values.

[0070] The system compares the actual change in cognitive performance (measured as 0%, i.e., the same as the pre-exercise baseline) with the predicted value of the cognitive impact of prescription A in step S4 by the system's digital twin (predicted as a slight decrease of 2%), and also compares the actual changes in inflammatory markers with the predicted values.

[0071] The comparison revealed that the log-soil axis model's prediction for this user was too conservative (overestimating the risk of log-soil crossover).

[0072] Furthermore, the comparison process employs the Bayesian update method to calculate the confidence interval of the prediction error, which serves as the basis for subsequent parameter adjustments.

[0073] S4: Model parameter iterative optimization, adaptive execution and evolution module. Based on the comparison results, it is determined that the log-soil axis (muscle-brain axis) sub-model may be slightly conservative in its prediction of this user under this type of exercise. The background optimization algorithm is started to fine-tune the internal parameters related to neuroinflammatory sensitivity in the sub-model (such as reducing the effect coefficient k1 of muscle factor X).

[0074] Once optimized, the updated system digital twin will be used for the user's future prescription generation and risk prediction, making the risk prediction more closely reflect the user's actual response.

[0075] Furthermore, parameter optimization employs an online gradient descent method to fine-tune the internal parameters of the five-dimensional Schenck network sub-model, aiming to minimize prediction error. The optimization process records model versions and supports rollback and A / B testing.

[0076] S5: Periodic Evolution. Through multiple iterations, the system's personalized five-dimensional network effect prediction for the user will become increasingly accurate, and the five-dimensional network balance early warning capability will continuously improve with usage time.

[0077] This embodiment achieves real-time security and user experience optimization through dual-loop control during execution; through feedback loop after execution, the system continuously optimizes the prediction accuracy of its core five-dimensional Sinker network sub-model using real cross-system functional data, thus realizing the system's self-evolutionary characteristics.

[0078] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

[0079] Although this paper frequently uses terms such as multi-source data management module, mind-body dual digital twin engine, prescription optimization generation module, simulation and deduction decision-making module, and adaptive execution and evolution module, the possibility of using other terms is not excluded. These terms are used merely for the convenience of describing and explaining the essence of this invention; interpreting them as any additional limitation would contradict the spirit of this invention.

Claims

1. A dynamic timing programming method for exercise prescriptions based on dual digital twins and a five-dimensional genotyping model, characterized in that, Includes the following steps: S1: Acquire multi-source health data of users and perform personalized initialization of pre-constructed system digital twins and cognitive digital twins based on the data; wherein, the system digital twin integrates a five-dimensional interaction network sub-model for simulating the interaction between different physiological dimensions. S2: Acquire macroscopic motion decision-making instructions and real-time monitoring data of the user, and fuse them with the current state data output by the system digital twin and cognitive digital twin to obtain the user's multi-dimensional comprehensive state; S3: Input the multidimensional integrated state into a multi-objective optimization model to generate one or more candidate exercise prescriptions; wherein, the constraint library of the multi-objective optimization model contains five-dimensional Sinker effect constraint values ​​calculated by the five-dimensional Sinker network sub-model based on the current multidimensional integrated state, which are used to quantitatively evaluate the potential negative impact of exercise on non-target physiological dimensions; S4: Input one or more candidate exercise prescriptions into the initialized system digital twin and cognitive digital twin respectively for collaborative simulation and deduction, and obtain the prediction results of the impact of each candidate prescription on multiple physiological dimensions and the prediction results of the impact on the user's psychological state; wherein, the simulation and deduction of the system digital twin includes the assessment of the activation degree of the five-dimensional Sinker network effect; based on a preset comprehensive evaluation function, select the final exercise prescription from the candidate exercise prescriptions, and the comprehensive evaluation function includes at least a penalty term for the risk of imbalance of the five-dimensional Sinker network; S5: During the user's execution of the final exercise prescription, the exercise parameters are dynamically adjusted based on real-time sensing data, and the parameters of the system's digital twin or multi-objective optimization model are iteratively optimized using multi-dimensional feedback data after the user's execution.

2. The method for dynamic timing programming of exercise prescriptions based on dual digital twins and a five-dimensional genotyping model according to claim 1, characterized in that, The five-dimensional Sinker effect constraint value is a neurocognitive inhibition risk value, which is calculated through the muscle-brain axis signal transduction sub-model in the system's digital twin and is used to quantitatively assess the potential negative impact of candidate exercise prescriptions on central nervous system function.

3. The method for dynamic timing programming of exercise prescriptions based on dual digital twins and a five-dimensional genotyping model according to claim 1, characterized in that, The five-dimensional Sinker effect constraint value is the risk value for immune homeostasis disturbance. It is calculated by the motion-immune axis signal transduction sub-model in the system's digital twin and is used to quantitatively assess the potential negative impact of candidate exercise prescriptions on the homeostasis of the body's immune system.

4. The method for dynamic timing programming of exercise prescriptions based on dual digital twins and a five-dimensional genotyping model according to claim 1, characterized in that, The multi-source health data in step S1 includes the user's genomic data; personalized initialization further includes: setting the initial sensitivity parameters of the corresponding five-dimensional genotyping network sub-model in the system's digital twin based on the gene locus information in the genomic data that is related to athletic ability, injury risk, and sensitivity to specific five-dimensional genotyping pathways.

5. The method for dynamic timing programming of exercise prescriptions based on dual digital twins and a five-dimensional sense-kinesthetic model according to claim 1, characterized in that, The constraint library for the multi-objective optimization model in step S3 also integrates one or more of the following constraints: Real-time load and recovery window constraints of the body based on dynamic metabolomics data assessment; Inflammation and nutritional status constraints based on microbiome data association; Psychological compliance constraints based on cognitive digital twin predictions; Priority protection dimension target constraints from the upper management system.

6. The method for dynamic timing programming of exercise prescriptions based on dual digital twins and a five-dimensional sense-kinesthetic model according to claim 1, characterized in that, The collaborative simulation and deduction of the system digital twin in step S4 is a multi-scale, integrated predictive simulation of molecular pathway activation, cellular stress response, tissue biomechanical changes, five-dimensional sense network signal transduction and its impact on the function of distal organ systems.

7. The method for dynamic timing programming of exercise prescriptions based on dual digital twins and a five-dimensional genotyping model according to claim 1, characterized in that, The multi-dimensional feedback data in step S5 includes the results of neurocognitive function tests or the detection values ​​of specific biomarkers; the iterative optimization specifically involves comparing the actual measured feedback data with the predicted values ​​of the system's digital twin in step S4, and calibrating the internal parameters of the relevant five-dimensional Schenck network sub-model based on the comparison deviation.

8. A dynamic timing programming system for exercise prescriptions based on dual digital twins and a five-dimensional Sinker model, used to implement the method according to any one of claims 1 to 7, characterized in that, include: The multi-source data management module is used to acquire, store, and process users' multi-source health data; The mind-body dual digital twin engine, connected to the multi-source data management module, is used to construct and initialize the system digital twin and the cognitive digital twin; among them, the system digital twin integrates a five-dimensional interaction network simulation sub-model for simulating the interaction between different physiological dimensions. The prescription optimization generation module is connected to the mind-body dual digital twin engine. Its built-in multi-objective optimization model integrates a constraint library containing five-dimensional generative and restraint effect constraint rules. The simulation and decision-making module is connected to the mind-body dual digital twin engine and the prescription optimization and generation module, respectively. It is used to drive the dual digital twin to perform collaborative simulation of candidate prescriptions and make safety decisions based on a comprehensive evaluation function that includes a negative score for the risk of imbalance in the five-dimensional Sinker network. The adaptive execution and evolution module, connected to the simulation and decision-making module, is used to adjust motion parameters based on real-time data during execution. It provides real-time visual guidance and voice / text prompts through a personalized interactive interface, collects subjective user feedback data, and gathers multi-dimensional feedback data after execution to optimize the system model.

9. A dynamic timing programming system for exercise prescriptions based on dual digital twins and a five-dimensional sense-kinesthetic model, as described in claim 8, is characterized in that... The system digital twin is a multiscale computable model that integrates a biomechanical model, a metabolic network model, an endocrine and immune model, and at least one five-dimensional seroconstriction network effect model. The five-dimensional seroconstriction network effect model includes a muscle-brain axis model or a motor-immune axis model.

10. A dynamic timing programming system for exercise prescriptions based on dual digital twins and a five-dimensional sense-kinesthetic model, as described in claim 8, is characterized in that... This system includes a data interface module, which is used to receive quantitative risk data reports from external health assessment services, or to feed back prescription execution results and twin status updates to the upper-level time-series decision engine.

11. A dynamic timing programming system for exercise prescriptions based on dual digital twins and a five-dimensional sense-kinesthetic model, as described in claim 8, is characterized in that... It also includes a personalized interaction and guidance interface module, which is connected to the adaptive execution and evolution module. This module is used to present users with executable exercise prescriptions packaged with guidance strategies, provide real-time visual guidance and voice / text prompts during execution, and collect users' subjective feedback data.

Images