A method for realizing exercise health management based on individualized cardiopulmonary function digital twin model

Abstract

A kind of individualized cardiorespiratory function digital twin model based on exercise health management method.The method obtains the cardiorespiratory exercise test (CPET) data of user as reference, inputs into cardiorespiratory function digital twin model, generates the cardiorespiratory function limit and sudden illness risk prediction for user, achieves the purpose of individualized exercise health management.The technical scheme adopted by the present application firstly uses partial differential equation set to describe the oxygen metabolism chain of respiratory-circulation-function organ, then uses digital twin neural network to fit the metabolism chain, inputs individual characteristics and exercise load parameters, and outputs oxygen metabolism results and internal function characteristic parameters of human body;Further, the output of digital twin network is input into the prediction network to realize the cardiovascular disease risk prediction.The personalized cardiorespiratory function digital twin model generated by the present application provides a powerful tool for more scientific and efficient physical training, early intervention and precise prevention of diseases.

Description

Technical Field

[0001] This invention belongs to the field of cardiopulmonary exercise testing. It designs an individualized digital twin model of cardiopulmonary function. By using data obtained by users through cardiopulmonary exercise tests, physiological mechanism models, and neural network algorithms, the digital twin model is constructed. The model simulates the individual user in terms of physiological function, generating a personalized digital twin model of cardiopulmonary function for the user. At the same time, it can effectively predict the individual's cardiopulmonary function limits and the risk of sudden acute illness. It can be used to carry out more scientific and efficient physical training, early intervention and precise prevention of diseases. Background Technology

[0002] Cardiopulmonary exercise testing is an objective, quantitative, and non-invasive examination method that includes a power treadmill, a fitted mask, a gas metabolism analyzer, and an electrocardiogram monitoring system. It can monitor key indicators throughout the entire process, from rest, warm-up, exercise, and recovery. However, current research, such as the study by Zhang Xiaohong's team published in the *Chinese Journal of Gerontology* titled "Construction of a Risk Prediction Model for Coronary Artery Stenosis Using Cardiopulmonary Exercise Testing," explores the relationship between cardiopulmonary exercise testing and coronary artery stenosis and establishes a risk prediction model. This model uses t-tests and logistic regression analysis to analyze potential risk factors affecting coronary artery stenosis, constructs a nomogram model, and internally validates it using the Bootstrap method for risk prediction. Another study, by Du Li's team published in the *Journal of Inner Mongolia Medical University* titled "Prediction of Postoperative Complications and Lung Function in Patients Undergoing Hepatobiliary Surgery Using Cardiopulmonary Exercise Testing," primarily focuses on patients undergoing hepatobiliary surgery. This study uses cardiopulmonary exercise test indicators to predict postoperative complications, applying normal distribution, t-tests, Wilcoxon rank-sum tests, and χ² tests. While professional statistical methods were used for comparative analysis, the selection of indicators was limited, mainly focusing on a few indicators such as VO2max / pred, VO2max / kg, AT, O2, and HR / pred, without fully utilizing the detailed data records provided by cardiopulmonary exercise testing equipment. Current research largely remains at the level of data processing, still requiring professional analysis by physicians for disease diagnosis. However, by leveraging machine learning and neural network technology, individualized digital twin models of cardiopulmonary function could be constructed, enabling automated machine assessment of physical condition and disease risk, providing assistance and suggestions for exercise and health management.

[0003] To construct a personalized digital twin model of cardiopulmonary function, this invention requires cardiopulmonary exercise test measurement data from subjects (such as metabolic equivalent, electrocardiogram, blood pressure, blood oxygen saturation, gas metabolism, respiratory rate, respiratory reserve, oxygen pulse, etc.), as well as corresponding health status diagnoses, including cardiopulmonary function limits (such as maximum heart rate, maximum oxygen uptake, forced vital capacity, etc.) and disease assessments (such as hypertension, valvular heart disease, pulmonary hypertension, pulmonary insufficiency, obesity, etc.). Using this data, more parameters (such as alveolar gas partial pressure, respiratory rhythm, neurally driven ventilation, pulmonary peripheral circulation flow, etc.) are first obtained through a physiological mechanism model. Then, combining all the data and indicators guides machine learning to obtain a digital twin model of cardiopulmonary function. Simultaneously, it can provide personalized probability assessments and risk classifications for potential diseases, offering individuals scientific health management plans and exercise prescription recommendations. Summary of the Invention

[0004] This invention proposes a method for sports health management based on a personalized digital twin model of cardiopulmonary function. Using the results of cardiopulmonary exercise tests of subjects as a reference, a digital twin model is constructed through a neural network to simulate the physiological functions of cardiopulmonary function under exercise conditions, generating a sports health management plan for individual subjects, thereby achieving the goal of intelligent individual diagnosis and health intervention.

[0005] The technical solution of the present invention:

[0006] A method for sports and health management based on a personalized digital twin model of cardiopulmonary function, comprising the following steps:

[0007] The first step is to construct a digital twin model of the physiological mechanisms of human cardiopulmonary function.

[0008] First, a physical model describing the physiological mechanism of human cardiopulmonary function is constructed. Then, the first neural network algorithm is used to learn the complex physical model to generate a lightweight digital model.

[0009] (1) The physical model of the physiological mechanism of human cardiopulmonary function includes the cardiovascular system, respiratory mechanics system, gas exchange system, cardiovascular control system, and respiratory control system.

[0010] Furthermore, the cardiovascular system is used to simulate blood flow within the circulatory system and the dynamic changes of the cardiovascular system during exercise. This includes a cardiac model, a pulmonary circulation model, a systemic circulation model, a vascular bed model, and a venous return model. The cardiac model includes the left and right atria and ventricles, considering blood flow through the mitral, aortic, tricuspid, and pulmonary valves; the pulmonary circulation model includes the pulmonary artery, peripheral vessels, and pulmonary veins; the systemic circulation model includes systemic arteries, peripheral vessels, systemic veins, and the vena cava; the vascular bed model is divided into vascular beds of agonist muscles and resting muscles, considering the local vasodilation mechanism during exercise; the venous return model includes a muscle pump and a respiratory pump, where the muscle pump refers to the effect of muscle contraction on venous return, and the respiratory pump refers to the effect of respiration on venous return. Preferably, the cardiovascular system includes a cardiac model, pulmonary circulation, systemic circulation, a respiratory pump, and a muscle pump.

[0011] Furthermore, the cardiovascular control system is used to simulate the regulation of the cardiovascular system by neural regulatory mechanisms. This includes a local blood flow rate regulation model, a central nervous system ischemic response model, afferent and efferent neural pathways, and a reflex control model of cardiovascular effectors. Specifically, the local blood flow rate regulation model simulates the vasodilatory response of the cerebral, coronary, and skeletal muscle vascular beds; the central nervous system ischemic response model simulates the effects of hypoxia on heart rate and systemic circulation; the afferent and efferent neural pathways simulate carotid sinus baroreceptors, peripheral chemoreceptors, and the pulmonary expansion reflex; and the reflex control model of cardiovascular effectors simulates the regulation of cardiac rhythm, ventricular contractility, peripheral vascular resistance, and venous capacity by the sympathetic and parasympathetic nervous systems. Cardiovascular control also includes the influence of respiratory central neuromuscular drives on efferent pathways and evaluates central command mechanisms as metabolic regulatory responses.

[0012] Furthermore, respiratory mechanics systems are used to simulate airflow generation and changes in lung volume during respiration, as well as adjustments in respiratory rate and tidal volume during exercise. This includes models representing the mechanical properties of the lungs and models representing the mechanical properties of the upper respiratory tract.

[0013] Furthermore, the respiratory control system utilizes information from central and peripheral chemoreceptors, as well as information from ventilation-related metabolic drivers, to estimate ventilation demand, thus forming a ventilation controller. Simultaneously, the respiratory controller also includes a respiratory pattern optimization method, which minimizes the work of breathing in each respiratory cycle by adjusting the respiratory pattern, thus forming a respiratory pattern optimizer.

[0014] Furthermore, the gas exchange system is used to simulate the transport and exchange of oxygen and carbon dioxide during exercise, as well as to maintain stable gas concentrations in the blood. This includes gas mixing and exchange models, gas transport models, and metabolic kinetic models. The gas mixing and exchange model considers alveolar gas mixing, blood gas exchange, and tissue-level gas exchange; the gas transport model simulates the transport and metabolism of oxygen and carbon dioxide in the brain and tissues; and the metabolic kinetic model regulates respiratory demand through metabolic rate ratios and metabolic-related neural drives. The cardiovascular system and respiratory mechanics interact through the gas exchange system to regulate blood gas concentrations.

[0015] Furthermore, the information interaction relationship between the physical models of the physiological mechanisms of human cardiopulmonary function is specifically as follows: For the cardiovascular control system: metabolic regulation requires external input of carbon dioxide and exhalation. The anaerobic threshold AT outputs the relative intensity of aerobic exercise I; the respiratory neuromuscular drive requires the ventilation volume VE output by the ventilation controller, the sensing time TI output by the breathing mode optimizer, and the blood flow rate BF, outputting the central respiratory neuromuscular drive response Nt; the efferent pathway requires the metabolic regulation output I, the respiratory neuromuscular drive output Nt, and the controlled variable θ output by the central nervous system ischemic response. aj and multiple input activities f of the input path output aj Output multiple outgoing activities f sj and the peak frequency f of the vagus nerve efferent fibers v Reflex-controlled effectors require metabolic regulation output I and efferent pathway output f. aj and f v The coronary artery region (HP), skeletal ischemic muscle region (RMP), active muscle region (AMP), and cerebral blood flow region (BP) are all areas where blood flow is locally controlled and output is controlled. jp The pressure of the smooth muscle extramuscular region (ep), spleen region (sp), coronary artery region (hp), skeletal resting muscle region (rmp), active muscle region (amp), and cerebral region (bp) is controlled by the local smooth muscle region (ep), spleen region (sp), coronary artery region (hp), skeletal resting muscle region (rmp), active muscle region (amp), and cerebral region (bp) fluid group (R). jp,t Heart rate (HR), elasticity of left and right ventricles at maximum systole (E) max,jv Left atrial blood flow Q la and peripheral skeletal resistance R amp,n The ischemic response of the central nervous system requires gas exchange and mixing of the brain's partial pressures of carbon dioxide (PaCO2) and oxygen (PaO2), with the output controlled variable θ. sj The afferent pathway requires gas exchange and mixing of outputs of PaCO2, PaO2, tidal volume (VT) from pulmonary mechanical output, and systemic output of P. SA Output f aj Local blood flow control requires the output of a reflex-controlled effector to measure the peripheral skeletal resistance R. amp,nThe metabolic regulation output I, the gas exchange and mixing output PaCO2, and the systemic circulation output inflow Q jp Output R jp .

[0016] For the cardiovascular system: systemic circulation requires the relative intensity of aerobic exercise (I) as the output of metabolic regulation and the circulating blood volume (V) as the output of effectors controlled by reflexes. u,jv The pressure R of each zone liquid group at time T of the effector output of the reflection control system. jp,t The extravascular pressure P of the active muscle vein output by the muscle pump im The abdominal pressure P output by the breathing pump abd The intrathoracic pressure P output by the breathing pump thor pulmonary arterial blood volume output by pulmonary circulation V pulmonary Q, the blood flow output by the heart to the left ventricle lv The right atrial pressure output by the heart, P ra and cardiac blood volume V heart Output right atrial blood flow Q ra Systemic arterial pressure P SA and inflow Q jp The heart requires a right atrial blood flow Q from the systemic circulation. ra The heart rate (HR) output by the reflex-controlled effector, and the elasticity of the left and right ventricles (E) at the maximum systolic moment output by the reflex-controlled effector. max,jv Q, the blood flow output from the left atrium via pulmonary circulation la and the P output of the breathing pump thor Output left ventricular blood flow Q lv Right atrial pressure P ra Cardiac blood volume V heart and right ventricular blood flow Q rv Pulmonary circulation requires a right ventricular blood flow Q from the heart. rv The intrathoracic pressure P output by the breathing pump thor Output of peripheral pulmonary blood flow Q pp pulmonary artery blood volume V pulmonary and left atrial blood flow Q la The breathing pump requires the sensing time (TI) output by the breathing mode optimizer, the tidal volume (VT) output by lung mechanics, and the blood flow rate (BF) output by the breathing mode optimizer, as well as the abdominal pressure (P). abd and intrathoracic pressure P thor The muscle pump requires no input; it outputs the extravascular pressure P of the active muscle vein. im .

[0017] For the respiratory control system: the mean detector requires the partial pressure of oxygen in the brain (PaO2) and the partial pressure of carbon dioxide in the brain (PaCO2) from gas exchange and mixing, and the pressure of carbon dioxide in cerebral venous blood (PbCO2) from gas transport, outputting the average PaO2 (PamO2), average PaCO2 (PamCO2), and average PbCO2 (PbmCO2); the ventilation controller requires the average PaO2 (PamO2), average PaCO2 (PamCO2), and average PbCO2 (PbmCO2) output from the mean detector, the metabolically related ventilation neural drive component (MRV) output from metabolic kinetics, and the blood flow rate (BF) output from the respiratory mode optimizer, outputting the tidal volume (VE); the respiratory mode optimizer requires the tidal volume (VE) output from the ventilation controller and the muscle pressure signal (P) output from lung mechanics. musc The output sensing time TI, blood flow rate BF, and neural drive response Nd are measured.

[0018] For gas exchange systems: gas exchange and mixing require pulmonary mechanical output of blood volume V and systemic circulatory inflow Q. jp Q, the blood flow output from the left atrium via pulmonary circulation la External input oxygen concentration (FiO2), external input carbon dioxide concentration (FiCO2), external input atmospheric pressure (P) atm pulmonary peripheral blood flow output from the pulmonary circulation Q pp The following parameters are considered in relation to pulmonary mechanical output: tidal volume (VT), mixed venous carbon dioxide concentration (CvCO2) and mixed venous oxygen concentration (CvO2), and arterial carbon dioxide concentration (CaCO2) and arterial oxygen concentration (CaO2). The following parameters are also considered in relation to gas exchange and mixing output: cerebral carbon dioxide partial pressure (PaCO2), arterial carbon dioxide concentration (CaCO2), arterial oxygen concentration (CaO2), and systemic circulatory inflow (Q). jp The system includes external input carbon dioxide uptake (VCO2), tidal volume (VT) of lung mechanical output, and external input oxygen uptake (VO2); output mixed venous carbon dioxide concentration (CvCO2), mixed venous oxygen concentration (CvO2), carbon dioxide metabolic production rate (MRTCO2), oxygen metabolic production rate (MRTO2), and cerebral venous blood carbon dioxide pressure (PbCO2); metabolic kinetics require gas transport output oxygen metabolic production rate (MRTO2) and gas transport output carbon dioxide metabolic production rate (MRTCO2); and output metabolically related ventilation neural drive components (MRV).

[0019] For respiratory dynamics systems: the upper respiratory tract requires pulmonary mechanical output of blood volume V and pulmonary mechanical output of pleural pressure P. pl Tidal volume VT and output airway flow gain G, which are related to lung mechanical output. AWLung mechanics requires an airway flow gain G from the upper respiratory tract output. AW External tracheal pressure P ao The neural drive response Nd output by the respiratory pattern optimizer, and the output blood volume V and pleural pressure P. pl Muscle pressure signal P musc And tidal volume VT.

[0020] The physical model of the physiological mechanism of human cardiopulmonary function is modeled and simulated using the Simulink library in MATLAB. First, a framework of five main modules is constructed by adding atomic subsystem modules. Considering the complexity of the formulas in the model, S-functions are used to write the formulas and parameters for each submodule, setting the inputs and outputs according to the model requirements and recording them in the corresponding S-Function modules. The input and output interfaces of each S-function block are then interconnected according to the parameter exchange in the model, thus completing the construction of the entire physiological model of human cardiopulmonary function.

[0021] (2) The first neural network algorithm adopts a sequence prediction model based on the Long Short-Term Memory (LSTM) network;

[0022] Input data includes: human parameters: muscle contraction duration, muscle contraction-relaxation cycle duration; directly measured quantities from cardiopulmonary exercise testing: VCO2, VO2, FiCO2, FiO2, AT; environmental inputs: P atm Pao; External environment: Temperature T (degrees Celsius), Humidity RH (percentage), Noise level L P (decibels); Exercise load: Load;

[0023] Output data includes: forced vital capacity, right atrial flow, left ventricular flow, right ventricular flow, intrathoracic pressure, peripheral pulmonary flow, cost minimization function, neurally driven ventilation, alveolar gas partial pressure, metabolic equivalent, respiratory exchange rate, load intensity, heart rate, heart rate reserve, oxygen pulse, stroke volume, systolic blood pressure, pulmonary artery diastolic blood pressure, blood oxygen saturation, oxygen uptake, oxygen uptake per kg of muscle, carbon dioxide excretion, minute ventilation, respiratory rate, respiratory reserve, oxygen ventilation equivalent, carbon dioxide ventilation equivalent, end-tidal oxygen partial pressure, and end-tidal carbon dioxide partial pressure. Furthermore, since physiological indicators change dynamically with exercise intensity, the input data is presented in time series form, specifically as a three-dimensional tensor, with dimensions defined as the number of samples (n), time step (Δt, i.e., observations at different time points), and number of features (y). i ), that is, (n,Δt,y) i To meet the processing requirements of neural networks, the input data needs to be normalized using the Z-score method.

[0024]

[0025] Among them, y i These are the original data points, μ is the mean of the data, and σ is the standard deviation of the data. The preprocessed dataset is divided into training and validation sets proportionally.

[0026] The first neural network algorithm implementation consists of two steps: a pre-training stage, which learns the general patterns of cardiopulmonary exercise from a large-scale population data to establish a basic framework; and a secondary training stage, which fine-tunes the network based on real individual data to optimize the network parameters so that they are precisely adapted to the characteristics of the individual, resulting in a personalized digital twin model.

[0027] Furthermore, the specific network training phases are as follows:

[0028] 1) Pre-training:

[0029] The input data comes from abnormal data of various indicators of normal people with random modifications, as well as the results of various indicators of subjects measured by cardiopulmonary exercise test (CPET);

[0030] The output data comes from the results of various indicators measured by some subjects through cardiopulmonary exercise test (CPET) and the internal functional parameters of the human body obtained by calculating the results of various indicators measured by other subjects through cardiopulmonary exercise test (CPET) in step (1) physiological mechanism model.

[0031] First, the network weights are initialized, and the corresponding parameters in the input dataset are used for forward propagation. The data passes through three LSTM hidden layers in sequence. In each hidden layer, the data is processed by weighted summation and non-linear activation functions. Finally, the predicted value is generated in the output layer.

[0032] The difference between the network's predicted values ​​and the values ​​generated from the simulated dataset is considered. The loss function, calculated using the mean squared error, is described as follows:

[0033]

[0034] Where L represents the loss function; y i This can represent the actual values ​​of the internal functional parameters of the human body calculated in the previous step, as well as the CPET test data; This represents the corresponding predicted value of the neural network (1st) pre-training; N refers to the number of output data of the neural network (1st).

[0035] The backpropagation process begins by calculating the gradient of the loss function relative to the network output. This is done layer by layer using the chain rule, calculating the gradient for each weight and bias. These gradients represent the contribution of each parameter to the final error. Next, the network weights are updated using stochastic gradient descent to reduce the loss, thus continuously adjusting the network's weight parameters to minimize the loss function. To prevent overfitting, a regularization technique (Dropout) is used after each hidden layer: during training, the outputs of some neurons in the network are randomly set to zero to avoid making the model too complex or overfitting the training data.

[0036] By iterating the forward and backward propagation processes repeatedly, the loss function tends to converge to a stable state.

[0037] 2) Second training:

[0038] The dataset for secondary training only requires the actual index results measured by the subjects through cardiopulmonary exercise testing (CPET) for input and output, and does not require the internal functional parameters of the human body calculated by the physiological mechanism model in the first step.

[0039] The process for secondary training is as follows:

[0040] First, the pre-trained network parameters are used as the initial weights. The input is the same as in the pre-training stage, and the predicted value is generated in the output layer through the same network structure.

[0041] Next, the loss function is used, and the CPET test index parameters are calculated from the mean square error according to formula (2) to account for the difference between the network prediction value and the actual measurement value. The backpropagation process is consistent with the pre-training stage. By minimizing the loss function, the network output can more accurately fit the individual test data, thus optimizing the network performance.

[0042] By iterating the forward and backward propagation processes repeatedly, the loss function tends to converge to a stable state.

[0043] The second step is to endow the digital twin model with sports and health management functions.

[0044] The second neural network still uses the LSTM method. Input data includes the output data from the first neural network during its second training; output data consists of the subjects' CPET diagnostic results. The preprocessed data is proportionally divided into training and validation sets. For the loss function, since disease prediction and health recommendations are performed simultaneously, separate loss functions need to be established and weighted before being combined into a final loss function.

[0045] The LSTM layer uses the sigmoid function to implement a gating mechanism, controlling the flow and forgetting of information; at the same time, the tanh function is used to perform nonlinear transformations on candidate memory cells and output values ​​to capture complex dependencies in the time series.

[0046] Furthermore, since physiological indicators change dynamically with exercise intensity, the input data is presented in time series form, specifically as a three-dimensional tensor. Its dimensions are defined as the number of samples (n), the time step (Δt, i.e., the observations at different time points), and the number of features (x). i ), that is, (n,Δt,x) i To meet the processing requirements of neural networks, the input data needs to be normalized using the Z-score method.

[0047]

[0048] Where, x i σ represents the original data points, μ is the mean of the data, and σ is the standard deviation of the data.

[0049] After training the neural network (2nd), the output needs to be processed by the softmax function:

[0050]

[0051] Among them, Soft(z) i It is the result of processing the i-th element in vector z using the Softmax function, thereby outputting the probability distribution of the diseases and exercise training suggestions according to the prescribed disease order.

[0052] The effects and benefits of this invention are: by utilizing relevant indicators and data provided by subjects through cardiopulmonary exercise experiments, the physiological mechanisms of cardiopulmonary function can be simulated, achieving the construction of an individualized digital twin model. This model aims to predict the limits of individual cardiopulmonary function and the risk of sudden acute illnesses, and to provide recommendations for exercise and health management. The digital twin model can serve as a virtual digital representation of the subject's own physiological mechanisms, and machine assessment can replace doctor's diagnosis. It can be used in areas such as conducting more scientific and efficient physical training, early intervention and precise prevention of diseases. Attached Figure Description

[0053] Figure 1 This is a flowchart illustrating the technical principle of the present invention.

[0054] Figure 2 This is a schematic diagram of a physiological model of human cardiopulmonary function.

[0055] Figure 3 These are internal schematics of a neural network; (a) is the first neural network; (b) is the second neural network. Detailed Implementation

[0056] To make the objectives and technical solutions of this invention clearer, the invention will be further described in detail below with reference to specific parameters and examples.

[0057] A method for sports and health management based on a personalized digital twin model of cardiopulmonary function, comprising the following steps:

[0058] The first step is to construct a digital twin model of the physiological mechanisms of human cardiopulmonary function.

[0059] The digital twin model of the physiological mechanism of human cardiopulmonary function is realized by integrating physical modeling and deep learning algorithms. It can be personalized to the cardiopulmonary function of subjects to update important indicator data.

[0060] The specific implementation can be divided into two processes. First, a physical model describing the physiological mechanism of human cardiopulmonary function is constructed. Then, the first neural network algorithm is used to learn the complex physical model to generate a lightweight digital model.

[0061] (1) The physical model of the physiological mechanism of human cardiopulmonary function is based on an integrated mathematical model, which includes five major modules: cardiovascular system, respiratory mechanics system, gas exchange system, cardiovascular controller, and respiratory controller. Each module contains different sub-modules. This model is based on previously validated cardiovascular and respiratory models and incorporates mechanisms that are dynamically related to aerobic exercise. It can predict the relevant parameters of cardiopulmonary function in healthy individuals during rest and aerobic exercise.

[0062] The five modules in the physical model of human cardiopulmonary function are described as follows:

[0063] Furthermore, the cardiovascular system is used to simulate blood flow within the circulatory system and the dynamic changes of the cardiovascular system during exercise. This includes a cardiac model, a pulmonary circulation model, a systemic circulation model, a vascular bed model, and a venous return model. The cardiac model includes the left and right atria and ventricles, considering blood flow through the mitral, aortic, tricuspid, and pulmonary valves; the pulmonary circulation model includes the pulmonary arteries, peripheral vessels, and pulmonary veins; the systemic circulation model includes systemic arteries, peripheral vessels, systemic veins, and the vena cava; the vascular bed model is divided into vascular beds of agonist muscles and resting muscles, considering the local vasodilation mechanism during exercise; the venous return model includes a muscle pump and a respiratory pump, where the muscle pump refers to the effect of muscle contraction on venous return, and the respiratory pump refers to the effect of respiration on venous return. Preferably, the cardiovascular system includes a cardiac model, pulmonary circulation, systemic circulation, a respiratory pump, and a muscle pump.

[0064] Furthermore, the Cardiovascular Controller is used to simulate the regulation of the cardiovascular system by neural regulatory mechanisms, particularly the regulation of heart rate and blood pressure during exercise. This includes models of local blood flow regulation, central nervous system ischemic response, afferent and efferent neural pathways, and reflex control models of cardiovascular effectors. Specifically, the local blood flow regulation model simulates the vasodilatory response of the cerebral, coronary, and skeletal muscle vascular beds; the central nervous system ischemic response model simulates the effects of hypoxia on heart rate and systemic circulation; the afferent and efferent neural pathways simulate carotid sinus baroreceptors, peripheral chemoreceptors, and the pulmonary expansion reflex; and the reflex control models of cardiovascular effectors simulate the regulation of cardiac rhythm, ventricular contractility, peripheral vascular resistance, and venous capacity by the sympathetic and parasympathetic nervous systems. Cardiovascular control also includes the influence of respiratory neuromuscular drives on efferent pathways and evaluates central command mechanisms as metabolic regulatory responses.

[0065] Furthermore, respiratory mechanics is used to simulate airflow generation and changes in lung volume during respiration, as well as adjustments in respiratory rate and tidal volume during exercise. This includes models representing the mechanical properties of the lungs and models representing the mechanical properties of the upper respiratory tract.

[0066] Furthermore, the respiratory controller utilizes information from central and peripheral chemoreceptors, as well as information from ventilation-related metabolic drivers, to estimate ventilation demand, thus forming a ventilation controller. Simultaneously, the respiratory controller also includes a breathing pattern optimization method, which minimizes the work of breathing in each respiratory cycle by adjusting the breathing pattern, thus forming a breathing pattern optimizer.

[0067] Furthermore, the gas exchange system is used to simulate the transport and exchange of oxygen and carbon dioxide during exercise, as well as maintaining stable gas concentrations in the blood. This includes gas mixing and exchange models, gas transport models, and metabolic kinetic models. The gas mixing and exchange model considers alveolar gas mixing, blood gas exchange, and tissue-level gas exchange; the gas transport model simulates the transport and metabolism of oxygen and carbon dioxide in the brain and tissues; and the metabolic kinetic model regulates respiratory demand through metabolic rate ratios and metabolic-related neural drives. The cardiovascular system and respiratory mechanics interact through the gas exchange system to regulate blood gas concentrations.

[0068] Furthermore, the information interaction relationship between the physical models of the physiological mechanisms of human cardiopulmonary function is specifically as follows: For the cardiovascular control system: metabolic regulation requires external input of carbon dioxide and exhalation. The anaerobic threshold AT outputs the relative intensity of aerobic exercise I; the respiratory neuromuscular drive requires the ventilation volume VE output by the ventilation controller, the sensing time TI output by the breathing mode optimizer, and the blood flow rate BF, outputting the central respiratory neuromuscular drive response Nt; the efferent pathway requires the metabolic regulation output I, the respiratory neuromuscular drive output Nt, and the controlled variable θ output by the central nervous system ischemic response. aj and multiple input activities f of the input path output aj Output multiple outgoing activities f sj and the peak frequency f of the vagus nerve efferent fibers v Reflex-controlled effectors require metabolic regulation output I and efferent pathway output f. aj and f v The coronary artery region (HP), skeletal ischemic muscle region (RMP), active muscle region (AMP), and cerebral blood flow region (BP) are all areas where blood flow is locally controlled and output is controlled. jp The pressure of the smooth muscle extramuscular region (ep), spleen region (sp), coronary artery region (hp), skeletal resting muscle region (rmp), active muscle region (amp), and cerebral region (bp) is controlled by the local smooth muscle region (ep), spleen region (sp), coronary artery region (hp), skeletal resting muscle region (rmp), active muscle region (amp), and cerebral region (bp) fluid group (R). jp,t Heart rate (HR), elasticity of left and right ventricles at maximum systole (E) max,jv Left atrial blood flow Q la and peripheral skeletal resistance R amp,n The ischemic response of the central nervous system requires gas exchange and mixing of the brain's partial pressures of carbon dioxide (PaCO2) and oxygen (PaO2), with the output controlled variable θ. sj The afferent pathway requires gas exchange and mixing of outputs of PaCO2, PaO2, tidal volume (VT) from pulmonary mechanical output, and systemic output of P. SA Output f aj Local blood flow control requires the output of a reflex-controlled effector to measure the peripheral skeletal resistance R. amp,n The metabolic regulation output I, the gas exchange and mixing output PaCO2, and the systemic circulation output inflow Q jp Output R jp .

[0069] For the cardiovascular system: systemic circulation requires the relative intensity of aerobic exercise (I) as the output of metabolic regulation and the circulating blood volume (V) as the output of effectors controlled by reflexes. u,jv The pressure R of each zone liquid group at time T of the effector output of the reflection control system. jp,t The extravascular pressure P of the active muscle vein output by the muscle pump im The abdominal pressure P output by the breathing pump abd The intrathoracic pressure P output by the breathing pump thor pulmonary arterial blood volume output by pulmonary circulation V pulmonary Q, the blood flow output by the heart to the left ventricle lvThe right atrial pressure output by the heart, P ra and cardiac blood volume V heart Output right atrial blood flow Q ra Systemic arterial pressure P SA and inflow Q jp The heart requires a right atrial blood flow Q from the systemic circulation. ra The heart rate (HR) output by the reflex-controlled effector, and the elasticity of the left and right ventricles (E) at the maximum systolic moment output by the reflex-controlled effector. max,jv Q, the blood flow output from the left atrium via pulmonary circulation la and the P output of the breathing pump thor Output left ventricular blood flow Q lv Right atrial pressure P ra Cardiac blood volume V heart and right ventricular blood flow Q rv Pulmonary circulation requires a right ventricular blood flow Q from the heart. rv The intrathoracic pressure P output by the breathing pump thor Output of peripheral pulmonary blood flow Q pp pulmonary artery blood volume V pulmonary and left atrial blood flow Q la The breathing pump requires the sensing time (TI) output by the breathing mode optimizer, the tidal volume (VT) output by lung mechanics, and the blood flow rate (BF) output by the breathing mode optimizer, as well as the abdominal pressure (P). abd and intrathoracic pressure P thor The muscle pump requires no input; it outputs the extravascular pressure P of the active muscle vein. im .

[0070] For the respiratory control system: the mean detector requires the partial pressure of oxygen in the brain (PaO2) and the partial pressure of carbon dioxide in the brain (PaCO2) from gas exchange and mixing, and the pressure of carbon dioxide in cerebral venous blood (PbCO2) from gas transport, outputting the average PaO2 (PamO2), average PaCO2 (PamCO2), and average PbCO2 (PbmCO2); the ventilation controller requires the average PaO2 (PamO2), average PaCO2 (PamCO2), and average PbCO2 (PbmCO2) output from the mean detector, the metabolically related ventilation neural drive component (MRV) output from metabolic kinetics, and the blood flow rate (BF) output from the respiratory mode optimizer, outputting the tidal volume (VE); the respiratory mode optimizer requires the tidal volume (VE) output from the ventilation controller and the muscle pressure signal (P) output from lung mechanics. musc The output sensing time TI, blood flow rate BF, and neural drive response Nd are measured.

[0071] For gas exchange systems: gas exchange and mixing require pulmonary mechanical output of blood volume V and systemic circulatory inflow Q. jpQ, the blood flow output from the left atrium via pulmonary circulation la External input oxygen concentration (FiO2), external input carbon dioxide concentration (FiCO2), external input atmospheric pressure (P) atm pulmonary peripheral blood flow output from the pulmonary circulation Q pp The following parameters are considered in relation to pulmonary mechanical output: tidal volume (VT), mixed venous carbon dioxide concentration (CvCO2) and mixed venous oxygen concentration (CvO2), and arterial carbon dioxide concentration (CaCO2) and arterial oxygen concentration (CaO2). The following parameters are also considered in relation to gas exchange and mixing output: cerebral carbon dioxide partial pressure (PaCO2), arterial carbon dioxide concentration (CaCO2), arterial oxygen concentration (CaO2), and systemic circulatory inflow (Q). jp The system includes external input carbon dioxide uptake (VCO2), tidal volume (VT) of lung mechanical output, and external input oxygen uptake (VO2); output mixed venous carbon dioxide concentration (CvCO2), mixed venous oxygen concentration (CvO2), carbon dioxide metabolic production rate (MRTCO2), oxygen metabolic production rate (MRTO2), and cerebral venous blood carbon dioxide pressure (PbCO2); metabolic kinetics require gas transport output oxygen metabolic production rate (MRTO2) and gas transport output carbon dioxide metabolic production rate (MRTCO2); and output metabolically related ventilation neural drive components (MRV).

[0072] For respiratory dynamics systems: the upper respiratory tract requires pulmonary mechanical output of blood volume V and pulmonary mechanical output of pleural pressure P. pl Tidal volume VT and output airway flow gain G, which are related to lung mechanical output. AW Lung mechanics requires an airway flow gain G from the upper respiratory tract output. AW External tracheal pressure P ao The neural drive response Nd output by the respiratory pattern optimizer, and the output blood volume V and pleural pressure P. pl Muscle pressure signal P musc And tidal volume VT.

[0073] The physical model of the physiological mechanism of human cardiopulmonary function is modeled and simulated using the Simulink library in MATLAB. First, a framework of five main modules is constructed by adding atomic subsystems. Considering the complexity of the formulas in the model, S-functions are used to write the formulas and parameters for each submodule, setting the inputs and outputs according to the model requirements and recording them in the corresponding S-Function modules. The input and output interfaces of each S-function block are then interconnected according to the parameter exchange in the model, thus completing the construction of the entire physiological model of human cardiopulmonary function.

[0074] The required parameters are input into Simulink to simulate the physiological model and calculate a series of internal human functional parameters; the results of various indicators measured by the subjects through cardiopulmonary exercise test (CPET) together provide the data basis for the neural network (1st).

[0075] The input parameters required to run the simulation are: directly measured quantities from the cardiopulmonary exercise test: VCO2 (ml), VO2 (ml), FiCO2 (%), FiO2 (%), AT (1 / min); environmental input quantities: P atm (mmHg), pao(cmH2O);

[0076] Table 1. List of pre-set parameters in the physiological model

[0077]

[0078] Table 2 List of parameters calculated in the physiological model

[0079]

[0080] Table 3. List of parameters measured during cardiopulmonary exercise testing

[0081]

[0082]

[0083] (2) The first neural network algorithm obtains a digital twin model by machine learning the physiological mechanism model of cardiopulmonary function, which transforms the model from complex to lightweight to improve computational efficiency.

[0084] This neural network has a fixed structure and uses a sequence prediction model based on a long short-term memory network (LSTM).

[0085] Input data includes:

[0086] Human parameters: muscle contraction duration, duration of muscle contraction-relaxation cycle;

[0087] Quantities directly measured during cardiopulmonary exercise testing: VCO2, VO2, FiCO2, FiO2, AT;

[0088] Environmental input: P atm Pao;

[0089] External environment: T, RH, L P ;

[0090] Exercise load: Load;

[0091] Output data includes: forced vital capacity, right atrial flow, left ventricular flow, left atrial flow, right ventricular flow, intrathoracic pressure, peripheral pulmonary circulation flow, cost minimization function, neurally driven ventilation, alveolar gas partial pressure, metabolic equivalent, respiratory exchange rate, workload, heart rate, heart rate reserve, oxygen pulse, stroke volume, systolic blood pressure, pulmonary artery diastolic blood pressure, blood oxygen saturation, oxygen uptake, oxygen uptake per kg of muscle, carbon dioxide excretion, minute ventilation, respiratory rate, respiratory reserve, oxygen ventilation equivalent, carbon dioxide ventilation equivalent, end-tidal oxygen partial pressure, and end-tidal carbon dioxide partial pressure.

[0092] Since physiological indicators change dynamically with exercise intensity, the input data is presented in time series form, specifically as a three-dimensional tensor. Its dimensions are defined as the number of samples (n), the time step (Δt, i.e., the observed values ​​at different time points), and the number of features (y). i ), that is, (n,Δt,y) i To meet the processing requirements of neural networks, the input data needs to be normalized using the Z-score method.

[0093]

[0094] Among them, y i Here are the original data points, μ is the mean, and σ is the standard deviation. After standardization, the data will have a distribution with a mean of 0 and a standard deviation of 1. This helps reduce the impact of outliers and speeds up convergence during model training. The preprocessed dataset is divided into training and validation sets in a 7:3 ratio.

[0095] The first neural network algorithm implementation consists of two steps: a pre-training stage, which learns the general patterns of cardiopulmonary exercise from a large-scale population data to establish a basic framework; and a secondary training stage, which fine-tunes the network based on real individual data to optimize the network parameters so that they are precisely adapted to the characteristics of the individual, resulting in a personalized digital twin model.

[0096] Furthermore, the specific network training phases are as follows:

[0097] 1) Pre-training: Before the model can perform personalized cardiopulmonary function simulation tasks, it needs to be trained on a large-scale dataset to learn the general patterns of cardiopulmonary function in a large population.

[0098] The pre-trained dataset needs to be generated by using the physical model of the physiological mechanism of human cardiopulmonary function in step (1) and randomly adjusting the preset constant parameters in the model within a reasonable range to generate a diverse simulated dataset that can represent most people.

[0099] The pre-training process is as follows:

[0100] The input data comes from the random modification of abnormal data of normal people’s indicators and the results of various indicators measured by subjects through cardiopulmonary exercise test (CPET);

[0101] The output data comes from the results of various indicators measured by some subjects through cardiopulmonary exercise test (CPET) and the internal functional parameters of the human body obtained by calculating the results of various indicators measured by other subjects through cardiopulmonary exercise test (CPET) in step (1) physiological mechanism model.

[0102] First, the network weights are initialized, and the corresponding parameters in the input dataset are used for forward propagation. The data passes through three LSTM hidden layers in sequence. In each hidden layer, the data is processed by weighted summation and non-linear activation functions. Finally, the predicted value is generated in the output layer.

[0103] Next, a loss function is used to measure the difference between the CPET test parameters of the subjects, the internal functional parameters of the human body calculated by the physical model of the physiological mechanism of human cardiopulmonary function in step (1), and the values ​​generated by the simulated dataset. The loss function is calculated using mean squared error and is described as follows:

[0104]

[0105] Where L represents the loss function; y i This can represent the actual values ​​of the internal functional parameters of the human body calculated in the previous step, as well as the CPET test data; This represents the corresponding predicted value of the neural network (1st) pre-training; N refers to the number of output data of the neural network (1st).

[0106] The smaller the loss value, the closer the model's prediction is to the actual data. By calculating the loss function, the network can understand the accuracy of its current prediction, providing a basis for the backpropagation process.

[0107] The backpropagation process begins by calculating the gradient of the loss function relative to the network output. This is done layer by layer using the chain rule, calculating the gradient for each weight and bias. These gradients represent the contribution of each parameter to the final error. Next, the network weights are updated using stochastic gradient descent to reduce the loss, thus continuously adjusting the network's weight parameters to minimize the loss function. To prevent overfitting, a regularization technique (Dropout) is used after each hidden layer: during training, the outputs of some neurons in the network are randomly set to zero to avoid making the model too complex or overfitting the training data.

[0108] By iteratively performing the forward and backward propagation processes, the loss function is brought to a stable convergence. This training result establishes the basic weight parameters in the neural network, directly reflecting the model's learning from a large-scale dataset, thus constructing a fundamental model capable of describing the general laws governing cardiopulmonary function.

[0109] 2) Secondary Training: During the pre-training phase, the network can initially grasp the general patterns of cardiopulmonary function in most people. However, due to differences in cardiopulmonary function among individuals, the current model cannot accurately simulate the physiological data of a specific individual. To achieve personalized fitting, the network must be trained a second time to construct a digital physiological model tailored to each individual subject.

[0110] The dataset for secondary training only requires the actual indicators measured by the subjects through the cardiopulmonary exercise test (CPET) for input and output, and does not need the internal functional parameters calculated by the physiological mechanism model in the first step. Although the output data in the training set is incomplete, this does not affect the model's goal of achieving individualized fitting. CPET test data can directly quantify the subjects' physical condition and reflect their individual physiological characteristics to a certain extent. Therefore, even without complete functional data, model fitting based solely on CPET data can still significantly improve the model's personalized accuracy.

[0111] The process for secondary training is as follows:

[0112] First, pre-trained network parameters are used as initial weights, with the same input as in the pre-training stage. These parameters are then processed through the same network structure to generate predicted values ​​at the output layer. In this process, the pre-training stage provides the network with initial capabilities, giving it a certain level of performance. Introducing individualized data provides a training foundation for specific individuals in the secondary training, thereby further improving the model's adaptability and accuracy.

[0113] Next, the loss function is used, and the CPET test index parameters are calculated from the mean square error according to formula (2) to account for the difference between the network prediction value and the actual measurement value. The backpropagation process is consistent with the pre-training stage. By minimizing the loss function, the network output can more accurately fit the individual test data, thus optimizing the network performance.

[0114] By iteratively performing the forward and backward propagation processes, the loss function eventually converges to a stable state. This training result effectively fits individualized data, allowing the weight parameters in the neural network to be optimally adjusted for each individual, thus achieving personalized model adaptation.

[0115] At this point, the pre-trained and second-trained individualized neural network (1st) can efficiently calculate the predicted values ​​of internal functional parameters of the human body, thus completely replacing the cardiopulmonary function physiological mechanism model. This significantly reduces the complexity of the model and greatly reduces the computational load. This output result, together with the subject's known CPET measurement data, is passed to the next neural network, achieving efficient and accurate determination of individualized cardiopulmonary function-related parameters.

[0116] The second step is to endow the digital twin model with sports and health management functions.

[0117] To further realize personalized disease risk assessment and exercise recommendations based on digital twin technology, a second neural network needs to be constructed. This network will utilize the output of the first personalized neural network, along with the subject's cardiopulmonary exercise test (CPET) data, to predict an individual's potential disease risk and provide personalized exercise recommendations. The basic training principle of the second neural network is essentially the same as the first; the training method for the second neural network is as follows:

[0118] The second neural network still uses the LSTM method. Input data includes the output data from the first neural network during its second training; output data consists of the subjects' CPET diagnostic results. The preprocessed data is divided into training and validation sets in a 7:3 ratio. For the loss function, since disease prediction and health recommendations are performed simultaneously, separate loss functions need to be established and weighted before being combined into a final loss function.

[0119] The LSTM layer uses the sigmoid function to implement a gating mechanism, controlling the flow and forgetting of information; at the same time, the tanh function is used to perform nonlinear transformations on candidate memory cells and output values ​​to capture complex dependencies in the time series.

[0120] Since physiological indicators change dynamically with exercise intensity, the input data is presented in time series form, specifically as a three-dimensional tensor. Its dimensions are defined as the number of samples (n), the time step (Δt, i.e., the observed values ​​at different time points), and the number of features (x). i ), that is, (n,Δt,x) i To meet the processing requirements of neural networks, the input data needs to be normalized using the Z-score method.

[0121]

[0122] Where, x iHere are the original data points, μ is the mean of the data, and σ is the standard deviation of the data. After standardization, the data will have a distribution with a mean of 0 and a standard deviation of 1. This helps to reduce the impact of outliers and speed up the convergence of the model during training. In addition, naming the predicted corresponding diseases with serial numbers helps to reduce the workload of the neural network.

[0123] After training the neural network (2nd), the output needs to be processed by the softmax function:

[0124]

[0125] Among them, Soft(z) i It is the result of processing the i-th element in vector z using the Softmax function, thereby outputting the probability distribution of the diseases and exercise training suggestions according to the prescribed disease order.

[0126] The second neural network (2nd) receives individual cardiopulmonary function-related parameters from the first neural network (1st) and generates predictions of individual cardiopulmonary function limits and the risk of sudden acute illness. This provides users with personalized, scientifically sound, and efficient physical training suggestions and assesses disease risks, thus enabling the digital twin model to function as a sports and health management system.

[0127] Table 4 List of Disease Risk Predictions

[0128]

[0129]

[0130]

[0131] Table 5. List of Exercise Training Recommendations

[0132]

[0133]

Claims

1. A method for sports health management based on a personalized digital twin model of cardiopulmonary function, characterized in that, The steps are as follows: The first step is to construct a digital twin model of the physiological mechanisms of human cardiopulmonary function. First, a physical model describing the physiological mechanism of human cardiopulmonary function is constructed. Then, the first neural network algorithm is used to learn the complex physical model to generate a lightweight digital model. (1) The physical model of the physiological mechanism of human cardiopulmonary function includes the cardiovascular system, respiratory mechanics system, gas exchange system, cardiovascular control system, and respiratory control system; The physical model of the physiological mechanism of human cardiopulmonary function is modeled and simulated using the Simulink library in MATLAB. First, five major module frameworks are constructed by adding atomic subsystem modules. Considering the complexity of the formulas in the model, S-functions are used to write the formulas and parameters in each sub-module, and the inputs and outputs are set according to the model requirements and recorded in the corresponding S-Function modules. The input and output interfaces of each S-function block are connected to each other according to the exchange of parameters in the model, thus realizing the construction of the entire physiological model of human cardiopulmonary function. (2) The first neural network algorithm adopts a sequence prediction model based on the Long Short-Term Memory (LSTM) network; The specific training stages of the first neural network are as follows: 1) Pre-training: First, the network weights are initialized, and the corresponding parameters from the input dataset are used for forward propagation. The data passes through three LSTM hidden layers in sequence. In each hidden layer, the data is processed by weighted summation and non-linear activation functions. Finally, the predicted value is generated in the output layer. The backpropagation process first requires calculating the gradient of the loss function relative to the network output, and then calculating the gradient of each weight and bias layer by layer using the chain rule. These gradients represent the contribution of each parameter to the final error; then, the network weights are updated using the stochastic gradient descent algorithm to reduce the loss, thereby continuously adjusting the network's weight parameters to minimize the loss function; to prevent overfitting, regularization is used after each hidden layer: during training, the outputs of some neurons in the network are randomly set to zero to avoid the model becoming too complex or overfitting the training data. By iterating the forward and backward propagation processes repeatedly, the loss function is brought to a stable convergence. 2) Secondary training: The process for secondary training is as follows: First, the pre-trained network parameters are used as the initial weights. The input is the same as in the pre-training stage, and the same network structure is used to generate the predicted value in the output layer. The backpropagation process is consistent with the pre-training phase; by minimizing the loss function, the network output can more accurately fit the individual's test data, thus optimizing network performance. By iterating the forward and backward propagation processes repeatedly, the loss function is brought to a stable convergence. The second step is to endow the digital twin model with sports and health management functions. The second neural network still uses the LSTM method; input data: including the output data of the first neural network used in the second training; output data: the CPET diagnosis results of the subjects; the preprocessed data is divided into training set and validation set according to the proportion; for the loss function, since disease prediction and health advice are performed simultaneously, loss functions need to be established separately and weighted to combine into the total loss function; The LSTM layer uses the sigmoid function to implement a gating mechanism, controlling the flow and forgetting of information; at the same time, the tanh function is used to perform nonlinear transformations on candidate memory cells and output values ​​to capture complex dependencies in the time series.

2. The method for sports health management based on a personalized cardiopulmonary function digital twin model according to claim 1, characterized in that, Since physiological indicators change dynamically with exercise intensity, the input data is presented in time series form, specifically as a three-dimensional tensor. Its dimensions are defined as the number of samples n, the time step Δt (i.e., the observed values ​​at different time points), and the number of features. To meet the processing requirements of neural networks, the input data needs to be normalized using the Z-score method. (1) ； in, These are the original data points. It is the mean of the data. It is the standard deviation of the data; the preprocessed dataset is divided into training and validation sets according to the proportions. In the pre-training process, the input data comes from abnormal data of the results of various indicators of normal people with random modifications, as well as the results of various indicators measured by the subjects through cardiopulmonary exercise test (CPET). The output data comes from the results of various indicators measured by the cardiopulmonary exercise test (CPET) of some subjects, as well as the internal functional parameters of the human body obtained by calculating the results of various indicators measured by the cardiopulmonary exercise test (CPET) of other subjects through the first step of the physiological mechanism model. For the difference between the network's predicted values ​​and the values ​​generated from the simulated dataset, the loss function is calculated using the mean squared error and described as follows: (2) ； in, Represents the loss function; This represents the actual values ​​of the internal functional parameters of the human body calculated in the previous step, as well as the CPET test data. This represents the corresponding predicted value of the first pre-trained neural network; This refers to the number of data points output by the first neural network. In the second training process, the input and output of the dataset for the second training only need the real index results measured by the subjects through the cardiopulmonary exercise test (CPET), and do not need the internal functional parameters of the human body calculated by the physiological mechanism model in the first step. Using the loss function, the CPET test index parameters are calculated from the mean square error according to formula (2) for the difference between the network prediction value and the actual measurement value; After training the second neural network, the output needs to be processed by the softmax function: (4) ； in, It is a vector The result of processing the i-th element in the data using the Softmax function is used to output the probability distribution of the diseases and exercise training suggestions according to the prescribed disease order. The cardiovascular system is used to simulate the flow of blood in the circulatory system and the dynamic changes of the cardiovascular system during exercise; it includes a heart model, a pulmonary circulation model, a systemic circulation model, a vascular bed model, and a venous return model. The heart model includes the left and right atria and ventricles, considering blood flow through the mitral, aortic, tricuspid, and pulmonary valves; the pulmonary circulation model includes the pulmonary artery, peripheral vessels, and pulmonary veins; the systemic circulation model includes systemic arteries, peripheral vessels, systemic veins, and the vena cava; the vascular bed model is divided into vascular beds of agonist muscles and resting muscles, considering the local vasodilation mechanism during exercise; the venous return model includes muscle pumps and respiratory pumps, where muscle pumps refer to the effect of muscle contraction on venous return, and respiratory pumps refer to the effect of respiration on venous return. The cardiovascular system includes a heart model, pulmonary circulation model, systemic circulation model, respiratory pump, and muscle pump.

3. The method for sports health management based on a personalized cardiopulmonary function digital twin model according to claim 1, characterized in that, In step (2), the input data includes: human parameters: muscle contraction duration, muscle contraction-relaxation cycle duration; and directly measured quantities from cardiopulmonary exercise testing: carbon dioxide excretion. Oxygen intake The proportion of carbon dioxide in inhaled gas The proportion of oxygen in the inhaled gas Anaerobic threshold (AT); Environmental input: atmospheric pressure airway pressure External environment: temperature ,humidity Noise level Exercise load: Output data includes: forced vital capacity, right atrial flow, left ventricular flow, right ventricular flow, intrathoracic pressure, peripheral pulmonary flow, cost minimization function, neurally driven ventilation, alveolar gas partial pressure, metabolic equivalent, respiratory exchange rate, workload, heart rate, heart rate reserve, oxygen pulse, stroke volume, systolic blood pressure, pulmonary artery diastolic blood pressure, blood oxygen saturation, oxygen uptake, oxygen uptake per kg of muscle, carbon dioxide excretion, minute ventilation, respiratory rate, respiratory reserve, oxygen ventilation equivalent, carbon dioxide ventilation equivalent, end-tidal oxygen partial pressure, and end-tidal carbon dioxide partial pressure. The cardiovascular control system is used to simulate the regulation of the cardiovascular system by neural regulatory mechanisms. It includes a local blood flow rate regulation model, a central nervous system ischemic response model, afferent and efferent neural pathways, and a reflex control model of cardiovascular effectors. Specifically, the local blood flow rate regulation model simulates the vasodilatory response of the cerebral, coronary, and skeletal muscle vascular beds; the central nervous system ischemic response model simulates the effects of hypoxia on heart rate and systemic circulation; the afferent and efferent neural pathways simulate carotid sinus baroreceptors, peripheral chemoreceptors, and the pulmonary expansion reflex; and the reflex control model of cardiovascular effectors simulates the regulation of cardiac rhythm, ventricular contractility, peripheral vascular resistance, and venous capacity by the sympathetic and parasympathetic nervous systems. The cardiovascular control also includes the influence of respiratory central neuromuscular drive on the efferent pathway and evaluates the central command mechanism as a metabolic regulatory response.

4. The method for sports health management based on a personalized cardiopulmonary function digital twin model according to claim 1, characterized in that, The respiratory mechanics system is used to simulate the generation of airflow and changes in lung volume during breathing, as well as the adjustment of respiratory rate and tidal volume during exercise; This includes models representing the mechanical properties of the lungs and models representing the mechanical properties of the upper respiratory tract.

5. A method for sports health management based on a personalized cardiopulmonary function digital twin model according to claim 1, characterized in that, The respiratory control system uses information from central chemoreceptors, peripheral chemoreceptors, and ventilation-related metabolic drivers to estimate ventilation demand, thus forming a ventilation controller. The respiratory controller also includes a respiratory pattern optimization method, which minimizes the work of breathing in each respiratory cycle by adjusting the respiratory pattern, thus forming a respiratory pattern optimizer.

6. The method for sports health management based on a personalized cardiopulmonary function digital twin model according to claim 1, characterized in that, The gas exchange system is used to simulate the transport and exchange of oxygen and carbon dioxide during exercise, as well as to maintain a stable concentration of gases in the blood. It includes gas mixing and exchange models, gas transport models, and metabolic kinetic models. Among them, the gas mixing and exchange model considers the mixing of alveolar gases, blood gas exchange, and tissue-level gas exchange; the gas transport model simulates the transport and metabolism of oxygen and carbon dioxide in the brain and tissues; the metabolic kinetic model regulates respiratory demand through metabolic rate ratios and metabolic-related neural drives; the cardiovascular system and respiratory mechanics interact through the gas exchange system to regulate the concentration of gases in the blood.

7. The method for sports health management based on a personalized cardiopulmonary function digital twin model according to claim 1, characterized in that, The information interaction relationship between the physical models of the physiological mechanisms of human cardiopulmonary function is as follows: For the cardiovascular control system: metabolic regulation requires external input of carbon dioxide and exhalation. and anaerobic threshold Output of relative intensity of aerobic exercise ; The respiratory system's neuromuscular drive requires the ventilation output from the ventilation controller. Sensing time output by the breathing mode optimizer And blood flow rate (BF), output of central respiratory neuromuscular drive response ; Efferent pathways require metabolic regulation of output. Respiratory system neuromuscular drive output Controlled variables of ischemic response output in the central nervous system Various input activities and input path outputs Output multiple outgoing activities and the peak frequency of vagal nerve efferent fibers Reflex-controlled effectors require metabolic regulation of output. Output of the transmission path and and the coronary artery region where blood flow is locally controlled. Skeletal resting muscle area Activation muscle area brain regions Liquid pressure The smooth muscle extra-area where blood flow is locally controlled. Spleen area Coronary artery area Skeletal resting muscle area Activation muscle area brain regions Liquid pressure Heart rate Elasticity of the left and right ventricles at the moment of maximum contraction Left atrial blood flow and peripheral skeletal resistance The ischemic response of the central nervous system requires gas exchange and the mixed output of cerebral carbon dioxide partial pressure. and partial pressure of oxygen in the brain Output controlled variables ; The input pathway requires gas exchange and mixing of output gases. , Tidal volume of lung mechanical output , body circulation output Output ; Local blood flow control requires the output of a reflex-controlled effector to provide peripheral skeletal resistance. metabolic regulatory output Gas exchange and mixed output Inflow of fluid circulation output Output ; For the cardiovascular system: systemic circulation requires a relatively high intensity of aerobic exercise that regulates metabolic output. Circulating blood volume output by the effector controlled by reflexes The pressure of each zone liquid group at time T of the effector output of the reflection control. Extravascular pressure of the active muscle veins output by the muscle pump Abdominal pressure output by the breathing pump Intrathoracic pressure output by the breathing pump pulmonary artery blood volume output by pulmonary circulation Left ventricular blood flow output by the heart Right atrial pressure output by the heart and the cardiac output volume Output right atrial blood flow Systemic arterial pressure and inflow The heart needs blood flow from the right atrium pumped out by the systemic circulation. Heart rate output by the reflex-controlled effector The maximum systolic elasticity of the left and right ventricles at the moment of maximum contraction of the effector output of the reflex-controlled system. Left atrial blood flow output from pulmonary circulation and the output of the breathing pump Output left ventricular blood flow Right atrial pressure Cardiac blood volume and right ventricular blood flow ; Pulmonary circulation requires blood flow from the right ventricle pumped by the heart. Intrathoracic pressure output by the breathing pump Experiencing increased peripheral pulmonary blood flow pulmonary artery blood volume and left atrial blood flow ; The breathing pump requires the sensing time output by the breathing mode optimizer. Tidal volume of lung mechanical output and the blood flow rate output by the breathing mode optimizer Output abdominal pressure and intrathoracic pressure The muscle pump requires no input; it outputs the extravascular pressure of the active muscle veins. ; For respiratory control systems: mean detectors require gas exchange and mixed output of brain oxygen partial pressure. partial pressure of carbon dioxide in the brain during gas exchange and mixing output And the pressure of carbon dioxide in cerebral venous blood transported out by gas. Output the average value of PaO2 PaCO2 average value and average PbCO2 The ventilation controller requires the average PaO2 output from the average detector. PaCO2 average value Average value of PbCO2 Metabolic kinetic output of metabolic-related ventilation neural drive components and the blood flow rate output by the breathing mode optimizer Output ventilation ; The breathing mode optimizer requires the ventilation output from the ventilation controller. and muscle pressure signals output by lung mechanics Output sensing time Blood flow rate and neural drive response ; For gas exchange systems: gas exchange and mixing require the blood volume produced by pulmonary mechanical output. Inflow of body circulation output Left atrial blood flow output from pulmonary circulation The concentration of inhaled oxygen from the outside The concentration of inhaled carbon dioxide from external sources External atmospheric pressure pulmonary peripheral blood flow output from the pulmonary circulation Tidal volume of lung mechanical output Mixed intravenous carbon dioxide concentration in gas transport output Mixed intravenous oxygen concentration with gas transport output arterial carbon dioxide concentration and arterial oxygen concentration Gas transport requires gas exchange and mixing, including the partial pressure of carbon dioxide output from the brain. Gas exchange and mixing output of arterial carbon dioxide concentration Gas exchange and mixing output arterial oxygen concentration Inflow of body circulation output Carbon dioxide intake from external sources Tidal volume of lung mechanical output Oxygen intake from the outside Output mixed intravenous carbon dioxide concentration Mixed intravenous oxygen concentration Carbon dioxide metabolic production rate Oxygen metabolism rate and the pressure of carbon dioxide in cerebral venous blood Metabolic kinetics requires the rate of oxygen production from gas transport output. Metabolic production rate of carbon dioxide exported by gas transport Output of ventilation-driven components related to metabolism ; For respiratory dynamics systems: the upper respiratory tract requires the blood volume of pulmonary mechanical output. pleural pressure output by lung mechanics Tidal volume of lung mechanical output Output airway flow gain ; Lung mechanics requires an increase in airway flow rate output from the upper respiratory tract. External tracheal pressure Neural drive response of breathing pattern optimizer output Output blood volume pleural pressure Muscle pressure signals and tidal volume .

8. A method for sports health management based on a personalized cardiopulmonary function digital twin model according to claim 1, characterized in that, In the second step: Since physiological indicators change dynamically with exercise intensity, the input data is presented in time series form, specifically a three-dimensional tensor, whose dimensions are defined as the number of samples n, the time step Δt, and the number of features. To meet the processing requirements of neural networks, the input data needs to be normalized using the Z-score method. (3) ； in, These are the original data points. It is the mean of the data. It is the standard deviation of the data.

Images