Methods and systems for evaluating the effectiveness of cardiac rehabilitation

Abstract

This invention discloses a method and system for evaluating the effectiveness of cardiac rehabilitation, relating to the field of medical and health technology. The method includes: acquiring physiological monitoring data, individualized parameter data, comorbidity data, multidrug therapy data, pharmacokinetic parameters, and pharmacodynamic interaction parameters of the target patient; generating baseline physiological index data without drug intervention based on the individualized parameter data and comorbidity data through a pre-set individualized cardiac response baseline model; calculating drug concentration based on multidrug therapy data and pharmacokinetic parameters; calculating comprehensive pharmacodynamic data by combining pharmacodynamic interaction parameters; generating predicted physiological data under multidrug intervention; subsequently obtaining purified physiological data through adaptive filtering and frequency domain decomposition; and finally generating a rehabilitation effect score through a pre-trained long short-term memory network model. Its beneficial effects include: distinguishing between drug interference and real cardiac function improvement, and improving the accuracy of rehabilitation assessment for patients undergoing multidrug therapy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical health, and particularly relates to a heart rehabilitation effect evaluation method and system. BACKGROUND

[0002] Heart rehabilitation is an important means for cardiovascular disease patients to recover heart function and improve life quality, and its core goal is to promote heart function improvement through progressive exercise training and life intervention. Accurate and real-time rehabilitation effect evaluation can not only guide the rehabilitation training intensity of patients, but also effectively reduce the risk of heart events and improve rehabilitation safety.

[0003] During the heart rehabilitation process, heart rate is a core index reflecting heart function state and autonomic nervous regulation ability. The existing technology usually evaluates the heart rehabilitation progress of patients through heart rate changes during exercise, resting and recovery heart rate indexes, and heart rate variability. These methods have been relatively mature in clinical application and can provide important reference for rehabilitation effect.

[0004] However, for patients in the medication period who are using heart drugs (such as beta blockers, calcium channel blockers, etc.), the heart rate of the patients is often affected by the drugs, showing a decrease or fluctuation in heart rate level. Since the drug effect has dynamicity, interaction and individual difference, the existing heart rate monitoring technology is difficult to distinguish the influence of the drugs on the heart rate from the real changes in the rehabilitation process itself, thereby affecting the accuracy of the rehabilitation effect evaluation.

[0005] Therefore, a heart rehabilitation effect evaluation method and system are proposed. SUMMARY

[0006] In view of the above existing technical conditions, the present application is proposed. Embodiments of the present application provide a heart rehabilitation effect evaluation method and system, which can distinguish drug interference from real heart function improvement and improve the rehabilitation evaluation accuracy of patients with multiple drug treatments.

[0007] According to an aspect of the present application, a method for evaluating the effect of cardiac rehabilitation is provided, comprising: obtaining physiological monitoring data, individualized parameter data, comorbidity data, multi-drug treatment data, and pharmacokinetic interaction parameters and pharmacodynamic interaction parameters between drugs used; generating baseline physiological indicator data representing the target patient under no drug intervention according to the individualized parameter data and the comorbidity data through a preset individualized cardiac response baseline model; calculating the concentration of each drug in the target patient's body according to the multi-drug treatment data and the pharmacokinetic interaction parameters through a preset pharmacokinetic model; calculating comprehensive pharmacodynamic data under the combined action of multi-drugs according to the concentration and the pharmacodynamic interaction parameters through a preset pharmacodynamic model; generating predicted physiological data representing the target patient under the combined intervention of multi-drugs according to the comprehensive pharmacodynamic data and the baseline physiological indicator data; obtaining pure physiological data after removing drug interference based on the predicted physiological data and the physiological monitoring data through adaptive filtering and frequency domain decomposition processing; generating a rehabilitation effect score representing the real improvement of the target patient's cardiac function after removing drug interference through a pre-trained long short-term memory network model based on the pure physiological data.

[0008] According to another aspect of the present application, a system for evaluating the effect of cardiac rehabilitation is provided, comprising: an acquisition module for obtaining physiological monitoring data, individualized parameter data, comorbidity data, multi-drug treatment data, and pharmacokinetic interaction parameters and pharmacodynamic interaction parameters between drugs used; an individual indicator generation module for generating baseline physiological indicator data representing the target patient under no drug intervention according to the individualized parameter data and the comorbidity data through a preset individualized cardiac response baseline model; a first calculation module for calculating the concentration of each drug in the target patient's body according to the multi-drug treatment data and the pharmacokinetic interaction parameters through a preset pharmacokinetic model; a second calculation module for calculating comprehensive pharmacodynamic data under the combined action of multi-drugs according to the concentration and the pharmacodynamic interaction parameters through a preset pharmacodynamic model; a prediction module for generating predicted physiological data representing the target patient under the combined intervention of multi-drugs according to the comprehensive pharmacodynamic data and the baseline physiological indicator data; an interference separation module for obtaining pure physiological data after removing drug interference based on the predicted physiological data and the physiological monitoring data through adaptive filtering and frequency domain decomposition processing; and an evaluation module for generating a rehabilitation effect score representing the real improvement of the target patient's cardiac function after removing drug interference through a pre-trained long short-term memory network model based on the pure physiological data.

[0009] According to another aspect of the present application, an electronic device is provided, comprising a memory for storing computer executable instructions, and a processor for executing the computer executable instructions, which, when executed by the processor, implement the steps of the method as described above.

[0010] According to another aspect of the present application, a computer storage medium is provided, having stored thereon computer executable instructions, which, when executed by a processor, implement the steps of the method as described above.

[0011] Compared with the prior art, the cardiac rehabilitation effect evaluation method and system according to the embodiments of the present application can separate drug interference signals and extract real cardiac function improvement data, and combine a long short-term memory network model for dynamic evaluation, thereby having the advantages of distinguishing between drug interference and real cardiac function improvement, and improving the rehabilitation evaluation precision of patients treated with multiple drugs. BRIEF DESCRIPTION OF DRAWINGS

[0012] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings in which:

[0013] Figure 1 Flowchart of the cardiac rehabilitation effect evaluation method of the present application.

[0014] Figure 2 Block diagram of the cardiac rehabilitation effect evaluation system of the present application.

[0015] Figure 3 Block diagram of an electronic device of the present application. DETAILED DESCRIPTION

[0016] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent to those skilled in the art that the described embodiments are merely a portion of the embodiments of the present application and thus are not limited to the described embodiments. In the drawings:

[0017] Exemplary method

[0018] In the traditional cardiac rehabilitation effect evaluation system, the dynamic interference of drug intervention on heart rate indicators has become a key factor affecting the accuracy of evaluation results. This interference makes it difficult for the evaluation results to truly reflect the patient's physiological state. Specifically, in the multi-drug treatment scenario, there are differences in drug metabolism kinetics, and the nonlinear superposition effect of drug efficacy interaction. Influenced by this, the heart rate variability, exercise tolerance and other key physiological parameters used to evaluate cardiac function cannot accurately reflect the actual improvement of cardiac function.

[0019] Moreover, drug competitive inhibition can also cause clearance fluctuations, which leads to deviations in blood drug concentration prediction, and further causes distortion of data after coupling baseline physiological indicators and drug effects. Taking cardiac rehabilitation patients using β-receptor blockers and calcium channel blockers as an example, the two drugs are competitively metabolized by the cytochrome enzyme system, which prolongs the half-life. In this way, the monitoring value of the heart rate recovery rate after exercise is mixed with the residual effect of the drug. The traditional frequency domain analysis method cannot effectively distinguish between the short-term heart rate oscillation caused by drug concentration fluctuations and the long-term trend signal of cardiac autonomic nervous function recovery, resulting in systematic noise interference in the data input of the rehabilitation score model.

[0020] If the above problems are not properly solved, drug dynamic interference will continue to pollute the physiological monitoring data stream. This will make the rehabilitation training intensity recommended value not match the actual cardiac load tolerance of the patient, increase the risk of exercise-induced arrhythmia. At the same time, evaluation bias will also hinder the dynamic optimization of individualized rehabilitation programs, prolong the patient's cardiac function recovery period, and even may mask the worsening trend of potential cardiac compensation mechanism, delay the best opportunity for clinical intervention.

[0021] In the face of the above problems, the present application first considers the influence mechanism of drug dynamic interference on physiological monitoring data, and finds that the nonlinear superposition effect of drug metabolism kinetics difference and drug efficacy interaction under multi-drug combination therapy is the core interference source. Therefore, the present application attempts to build a coupling model of drug action and cardiac physiological response, and separates the drug effect and real physiological change by simulating the predicted physiological data under multi-drug intervention. Further, the present application explores the use of frequency domain decomposition technology to distinguish between short-term signals caused by drug concentration fluctuations and long-term trends of cardiac function recovery, and introduces time series modeling methods to capture potential patterns of rehabilitation improvement.

[0022] Figure 1The method for evaluating the rehabilitation effect of the heart according to the embodiment of the present application is illustrated, which comprises: acquiring physiological monitoring data, individualized parameter data, comorbidity data, multi-drug treatment data and pharmacokinetic interaction parameters and pharmacodynamic interaction parameters of the drugs used; generating baseline physiological index data representing the target patient under no drug intervention according to the individualized parameter data and the comorbidity data through a preset individualized heart response baseline model; calculating the concentration of each drug in the target patient's body according to the multi-drug treatment data and the pharmacokinetic interaction parameters through a preset pharmacokinetic model; calculating comprehensive pharmacodynamic data under the combined action of multi-drugs according to the concentration and the pharmacodynamic interaction parameters through a preset pharmacodynamic model; generating predicted physiological data representing the target patient under the combined intervention of multi-drugs according to the comprehensive pharmacodynamic data and the baseline physiological index data; obtaining pure physiological data after removing the drug interference based on the predicted physiological data and the physiological monitoring data through adaptive filtering and frequency domain decomposition processing; and generating a rehabilitation effect score representing the real improvement of the heart function of the target patient after removing the drug interference through a pre-trained long short-term memory network model based on the pure physiological data.

[0023] The physiological monitoring data refers to real-time physiological signals such as heart rate, blood pressure and blood oxygen saturation collected by wearable devices or medical instruments, which can be realized by dynamic electrocardiogram monitors, smart bracelets or implantable sensors, and is used to reflect the current heart function state of the patient.

[0024] The individualized parameter data refers to individualized physiological indicators such as the age, gender, body fat rate, basal metabolic rate and cardiac ejection fraction of the patient, which can be obtained through electronic health records or clinical test reports, and is used to construct a patient-specific heart response model.

[0025] The comorbidity data refers to the diagnosis information of other diseases such as hypertension, diabetes and chronic kidney disease coexisting with the cardiovascular disease of the patient, which can be standardized recorded by using the international disease classification code, and is used to correct the pathological interference factors of the heart function baseline model.

[0026] The pharmacokinetic interaction parameters refer to the rate constant and half-life data of the absorption, distribution, metabolism and excretion process of the drug in the body, which can be obtained through in vitro experiments or population pharmacokinetic models, and are used to calculate the dynamic changes of the in vivo concentration when multi-drugs are used together.

[0027] The pharmacodynamic interaction parameters refer to the synergistic or antagonistic coefficients of different drugs in terms of target binding, signal pathway regulation or physiological effect superposition, which can be determined by in vitro cell experiments or clinical pharmacodynamics studies, and are used to quantify the comprehensive effect of multi-drug combination.

[0028] The individualized heart response baseline model refers to a mathematical simulation model established based on physiological characteristics and pathological states of a patient, and can be specifically trained by using a multiple linear regression or a machine learning algorithm to predict a benchmark index of heart function without drug intervention.

[0029] The core innovation of the present application is that the real physiological improvement signal of heart rehabilitation is extracted under the condition of complex drug interference, combined with individualized baseline prediction and dynamic pharmacokinetic calculation, to realize quantitative evaluation of rehabilitation effect of patients during medication.

[0030] As a preferred embodiment, the scheme of the present application is implemented as follows:

[0031] Firstly, physiological monitoring data of the target patient is obtained through the electronic medical record system of the hospital and the wearable device, including heart rate, blood pressure, electrocardiogram and other indicators. Individualized parameter data such as age, gender, weight, etc. is extracted from the patient's file. Comorbidity data includes related disease information such as hypertension, diabetes, etc. Multi-drug treatment data records the types, doses and medication times of drugs being taken by the patient. Pharmacokinetic interaction parameters and pharmacodynamic interaction parameters between drugs are provided by drug instructions and clinical pharmacology database.

[0032] Then, an individualized heart response baseline model based on physiological model is used to input the individualized parameters and comorbidity data of the patient, to simulate and generate baseline physiological index data of the patient without drug intervention, including resting heart rate, blood pressure and other indicators.

[0033] Then, a population pharmacokinetic model is used to calculate the concentration curves of each drug in the body over time, combined with the multi-drug treatment data and pharmacokinetic interaction parameters of the patient. The model takes into account the influence of drug-drug interactions on clearance rate.

[0034] Further, a pharmacodynamic model based on receptor occupancy theory is used to calculate the comprehensive pharmacodynamic data under the combined action of multiple drugs according to the concentration curves of each drug and the pharmacodynamic interaction parameters. The model takes into account the complex interactions such as synergy and antagonism between drugs.

[0035] Based on the comprehensive pharmacodynamic data and baseline physiological index data, the predicted physiological data is generated by using a pharmacodynamic-physiological response coupling model to simulate the changes of physiological indicators of the patient under the intervention of multiple drugs.

[0036] Subsequently, the predicted physiological data is compared with the actually monitored physiological data, the observation noise is removed by using an adaptive Kalman filtering algorithm, and then the low-frequency component representing the long-term rehabilitation trend is extracted by using wavelet transform for frequency domain decomposition, to obtain the pure physiological data without drug interference.

[0037] Finally, the purified physiological data is input into a pre-trained long short-term memory network model, which can capture the temporal features of the rehabilitation process, output physiological improvement indicators in multiple dimensions, and calculate the final rehabilitation effect score through a weighted scoring function.

[0038] Through the above-described scheme, this application can separate drug effects from actual physiological changes, improving the accuracy of cardiac rehabilitation effect assessment. This method considers the complex effects of multi-drug combination therapy, overcoming biases caused by drug interference in traditional assessment methods. By establishing an individualized baseline model and a drug-efficacy-physiological response coupling model, the simulation of patient-specific responses is achieved. The application of adaptive filtering and frequency domain decomposition techniques removes interference from short-term drug concentration fluctuations, extracting long-term trend signals reflecting cardiac function recovery. The introduction of a long short-term memory network can capture complex temporal patterns during the rehabilitation process, providing a more comprehensive basis for assessment. This multi-model collaborative assessment method not only improves the reliability of rehabilitation effect assessment but also provides important support for the development and dynamic adjustment of personalized rehabilitation programs, helping to optimize rehabilitation training intensity, reduce the risk of cardiac events, and improve rehabilitation safety and effectiveness.

[0039] In some of the schemes described above in this application, the effective concentration of each drug in the target patient is calculated using a pre-defined pharmacokinetic model. Specifically, this includes: extracting the competitive inhibitory relationships between the drugs based on multidrug therapy data, which are generated through cytochrome enzyme systems, plasma protein binding sites, and renal clearance pathways; calculating the net clearance rate of each drug in the case of multidrug coexistence based on the competitive inhibitory relationships and pharmacokinetic parameters; and calculating the effective concentration of each drug in the target patient over time based on the net clearance rate.

[0040] Among them, the cytochrome enzyme system involves the competition of oxidative metabolism mediated by isoenzymes such as CYP3A4 and CYP2D6, the plasma protein binding sites include the competition for the occupancy of binding sites such as albumin and α1-acid glycoprotein, and the renal clearance pathway covers the competition between glomerular filtration rate and renal tubular secretion channels.

[0041] The net clearance rate is calculated using a nonlinear kinetic model that integrates drug inhibition constants and enzyme activity parameters. For example, when the inhibition constant Ki of drug A on CYP3A4 is 5 μM, the metabolic clearance rate of drug B will decrease proportionally.

[0042] The concentration calculation is achieved through a set of differential equations. The inputs are the patient's weight, liver and kidney function parameters, and the drug's half-life. The output is a blood drug concentration curve that changes over time.

[0043] Specifically:

[0044] In the cytochrome enzyme system, if both drug X and drug Y are metabolized by CYP3A4, the competition coefficient will be calculated based on their enzyme affinity. For example, if the enzyme affinity of drug X is Km=10μM and that of drug Y is Km=15μM, then the competition coefficient is the reciprocal of the Km ratio.

[0045] In the plasma protein binding process, if drug Z occupies 80% of the albumin binding sites, the binding rate of drug W will be dynamically adjusted according to the remaining binding capacity.

[0046] When calculating net clearance, the original clearance of the drug can be multiplied by an inhibition factor determined by the competitive inhibition relationship. This inhibition factor is determined by the inhibition intensity matrix corresponding to the competitive inhibition relationship.

[0047] Ultimately, the effective concentration was obtained by solving the pharmacokinetic differential equation that included net clearance, administered dose, and patient physiological parameters. For simplicity, when using a one-compartment model, the expression for the decay of plasma concentration over time is: ,in The initial blood drug concentration is represented by the ratio of the initial drug dose to the volume of distribution, and k represents the elimination rate constant, which is the ratio of net clearance to the volume of distribution. For the more general case of multiple dosing, the corresponding differential equations need to be solved numerically.

[0048] By accurately quantifying the competitive effects of drugs on key metabolic and clearance pathways using the above methods, this approach can accurately predict the actual in vivo exposure levels of each drug when used in combination, providing a reliable input basis for subsequent comprehensive pharmacodynamic analysis.

[0049] In some of the schemes described above in this application, the comprehensive pharmacodynamic data under the combined action of multiple drugs is calculated through a preset pharmacodynamic model. Specifically, this includes: calculating the independent pharmacodynamic value of each drug on cardiac function based on the effective concentration of each drug and the preset maximum pharmacodynamic parameter and preset half-efficacy concentration parameter of each drug; determining the interaction relationship between any two drugs based on the pharmacodynamic interaction parameter, including synergistic effect, additive effect and antagonistic effect; calculating the combined pharmacodynamic value of the interaction between each pair of drugs based on the interaction relationship and the independent pharmacodynamic value; and integrating the pharmacodynamic values ​​after all drug interactions to obtain the comprehensive pharmacodynamic data under the combined action of multiple drugs.

[0050] The independent efficacy value is calculated using the Hill equation, with the maximum efficacy parameter corresponding to the maximum effect value of the drug and the half-efficacy concentration parameter corresponding to the drug concentration that produces 50% of the maximum effect.

[0051] The interaction relationship determination is based on the Loewe additive model or the Bliss independent model, and the interaction type is determined by comparing the difference between the theoretical additive effect and the actual observed effect.

[0052] The combined efficacy value is calculated using a nonlinear superposition method. For example, in one specific implementation:

[0053] when , When two drugs are determined to have a synergistic effect, their combined efficacy for:

[0054] ;

[0055] in, For the synergy coefficient, This is the parameter for maximum efficacy.

[0056] When determined to be an antagonistic effect, its combined efficacy for:

[0057] ;

[0058] in, The coefficient represents the antagonism coefficient.

[0059] In this process, priority rules are set for the integration of comprehensive pharmacodynamic data. As a preferred embodiment, for multiple drugs acting on the same cardiac receptor target, the target occupancy of each drug is calculated using a receptor occupancy model, and the pharmacodynamics are normalized and integrated based on their intrinsic activity, rather than simply adding them together.

[0060] Specifically, for multiple drugs targeting the same receptor, the drug effect with the highest intrinsic activity or receptor occupancy can be prioritized as the dominant effect, while the remaining drug effects are calculated and superimposed according to their receptor occupancy and intrinsic activity. By establishing a drug action network topology with drugs as nodes and interactions as edges, a graph traversal algorithm is used to process the interactions of all drug combinations, ultimately generating a comprehensive pharmacodynamic matrix. Each element of this matrix represents the net influence intensity of a specific drug combination on a specific cardiac function indicator.

[0061] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0062] Based on the effective concentration of each drug and its preset maximum efficacy parameter and preset half-effective concentration parameter, the independent efficacy value of each drug on cardiac function is calculated. For example, for the beta-blocker metoprolol, its maximum efficacy parameter is an 80% decrease in heart rate, and its half-effective concentration is 50 ng / mL. When the effective concentration of metoprolol is 100 ng / mL, its independent efficacy value calculated using the Hill equation is a 67% decrease in heart rate.

[0063] Based on drug-effect interaction parameters, the interaction relationship between any two drugs is determined, including synergistic, additive, and antagonistic effects. Specifically, by consulting drug interaction databases, a synergistic effect was determined between metoprolol and the calcium channel blocker verapamil; their combined action enhances the heart rate-lowering effect.

[0064] Based on the interaction relationships and independent efficacy values, the combined efficacy values ​​of drug interactions are calculated. Finally, the efficacy values ​​of all drug interactions are integrated to obtain the comprehensive efficacy data of multidrug combination. Thus, the combined efficacy of metoprolol and verapamil is integrated with the efficacy of other cardiac drugs to obtain the comprehensive efficacy data of multidrug combination, such as the percentage decrease in heart rate and the magnitude of blood pressure reduction.

[0065] Through the above technical solution, this application can calculate the comprehensive efficacy of multiple cardiac drugs used in combination, effectively taking into account the interactions between drugs and avoiding errors that may be caused by simply summing the individual efficacy of each drug. This method can more accurately predict the cardiac physiological state of patients under multidrug therapy, providing a reliable basis for subsequent removal of drug interference and evaluation of the true rehabilitation effect.

[0066] In some of the above-mentioned schemes of this application, pure physiological data after removing drug interference is obtained through adaptive filtering and frequency domain decomposition processing. Specifically, this includes: performing state estimation based on predicted physiological data and physiological monitoring data using an adaptive Kalman filter algorithm to remove observation noise and generate filtered physiological data; performing wavelet transform on the filtered physiological data to extract signal components within a preset frequency bandwidth corresponding to the long-term trend of cardiac rehabilitation improvement as pure physiological data.

[0067] The adaptive Kalman filter algorithm dynamically adjusts the process noise covariance matrix and the observation noise covariance matrix to match the time-varying characteristics of the patient's physiological signals in real time. For example, it increases the process noise covariance during the recovery phase after exercise to adapt to rapid changes in heart rate. Wavelet transform uses the sym4 wavelet basis function for five-level decomposition, breaking down the filtered physiological data into sub-signals of different frequency bands. The low-frequency component (0.04-0.15Hz) is selected as the long-term trend signal for the improvement of cardiac autonomic nervous system regulation. Kalman filtering and wavelet transform form a cascaded processing structure; the former eliminates high-frequency observation noise, while the latter separates low-frequency physiological trends, creating a complementary processing mechanism in the time and frequency domains.

[0068] Specifically, predicted physiological data and physiological monitoring data are synchronously input into an adaptive Kalman filter. A state equation describes the dynamic changes in cardiac physiological parameters, while an observation equation establishes the correlation between the predicted and measured data. The filter adjusts the Kalman gain coefficient in real time based on the information covariance matrix. When drug effects cause an increase in the deviation between predicted and measured data, the gain coefficient is automatically reduced to suppress abnormal interference. The filtered physiological data is decomposed into approximation coefficients and detail coefficients using wavelet transform. A soft thresholding method is used to remove residual noise in the high-frequency detail coefficients while retaining the low-frequency trend components in the approximation coefficients. The preset frequency bandwidth of 0.04-0.15Hz corresponds to the regulatory cycle of cardiac sympathetic and parasympathetic nerve activity, effectively characterizing the trend of improved heart rate variability brought about by rehabilitation training. This dual processing mechanism ensures that the pure physiological data retains the true signal of improved cardiac function while eliminating short-term interference caused by fluctuations in drug metabolism.

[0069] Through the above technical solution, this application can effectively remove short-term fluctuations and noise interference from physiological monitoring data, extracting pure physiological signals that reflect the long-term improvement trend of cardiac function. This allows for a more accurate assessment of the patient's true rehabilitation effect, avoiding the influence of drug interference and short-term fluctuations on the assessment results. Furthermore, this method can adapt to individual differences among different patients, improving the universality and reliability of rehabilitation effect assessment.

[0070] In some of the solutions described above in this application, a rehabilitation effect score is generated through a pre-trained long short-term memory network model. Specifically, this includes: acquiring the target patient's exercise history data; combining the pure physiological data, exercise history data, and individualized parameter data to form a time-series feature vector; inputting the time-series feature vector into the pre-trained long short-term memory network model, which identifies the temporal pattern of rehabilitation improvement and outputs multiple-dimensional physiological improvement indicators, including resting heart rate improvement value, post-exercise heart rate recovery improvement value, heart rate variability improvement value, exercise tolerance improvement value, and symptom improvement score; and calculating the rehabilitation effect score based on the resting heart rate improvement value, post-exercise heart rate recovery improvement value, heart rate variability improvement value, exercise tolerance improvement value, and symptom improvement score using a preset weighted scoring function.

[0071] The exercise history data includes the distribution of exercise types, durations, and intensities at different rehabilitation stages; the time series feature vector is synchronized with physiological data, exercise data, and individualized parameters according to a preset time window to form a multidimensional time series; the long short-term memory network model captures the dynamic correlation between heart rate improvement and exercise training through multiple hidden units; the weight coefficients in the weighted scoring function are determined based on clinical research data, for example, the weight coefficient for resting heart rate improvement is 0.3, and the weight coefficient for exercise tolerance improvement is 0.25.

[0072] Specifically, exercise history data is collected via wearable devices and aligned with clean physiological data by timestamps. This data, along with individualized parameters, forms the three dimensions of the feature vector. During the training phase, the Long Short-Term Memory (LSTM) network model uses historical rehabilitation case data to learn the delayed effects of different exercise patterns on heart rate variability improvement. The model's output layer has five independent neurons, corresponding to resting heart rate improvement, post-exercise heart rate recovery improvement, heart rate variability improvement, exercise tolerance improvement, and symptom improvement score. The weighted scoring function is a weighted sum of all indicators, with weight coefficients pre-set based on clinical experience.

[0073] Through the aforementioned technical solution, this application comprehensively considers the patient's physiological data, exercise status, and individual characteristics. It captures temporal patterns during the rehabilitation process using a long short-term memory network, assessing the improvement of cardiac function from multiple dimensions. A weighted scoring function further integrates various indicators, providing a quantitative score of rehabilitation effectiveness. This method avoids the limitations of single indicators and improves the comprehensiveness and accuracy of the assessment. Furthermore, by evaluating pure physiological data after removing drug interference, the influence of drug effects is eliminated, more accurately reflecting the true improvement of the patient's cardiac function.

[0074] In some of the schemes described above in this application, the preset weighted scoring function is: ;in, This represents the improvement in resting heart rate. This represents the improvement in heart rate recovery after exercise. This represents the improvement in heart rate variability. For improvement in exercise endurance, Assess symptom improvement scores. , , , and The preset weighting coefficients satisfy... .

[0075] The weighting coefficients are dynamically adjusted using clinical validation data, for example, for patients with coronary heart disease. , , , and The initial weights were 0.25, 0.20, 0.15, 0.30, and 0.10, respectively, while for patients with heart failure, they were adjusted to 0.15, 0.10, 0.20, 0.25, and 0.30. Each weight coefficient was determined using principal component analysis to ensure that the contribution of each dimension to the final score was positively correlated with its clinical importance. The constraint that the sum of the weights equals 1 was implemented using the Lagrange multiplier method to prevent score range drift due to improper weight allocation.

[0076] Specifically, the improvement in resting heart rate The improvement in heart rate recovery after exercise is calculated by the difference between the current resting heart rate and the baseline value. The improvement in heart rate variability was calculated using the logarithmic transformation of the rate of heart rate decline 3 minutes after exercise cessation. Based on the standardized results of the SDNN index, the improvement value of exercise endurance. Symptom improvement score quantified by percentage change in metabolic equivalents A five-point Likert scale was used for transformation. Each indicator was standardized using the Z-score and then multiplied by its corresponding weighting coefficient for linear combination.

[0077] For example, the improvement values ​​of various indicators of a patient with coronary heart disease after recovery are: =8 times / minute c = 15 times / minute =10ms =50 meters =Level 1 , , , and The values ​​are 0.25, 0.20, 0.15, 0.30, and 0.10 respectively. Substituting these values ​​into the function, we get: C = 0.25 × 8 + 0.2 × 15 + 0.15 × 10 + 0.3 × 50 + 0.1 × 1 = 21.6 points.

[0078] Through the above technical solution, this application can comprehensively consider multiple dimensions of physiological improvement indicators and calculate a quantitative rehabilitation effect score through a preset weighted scoring function. This method can objectively reflect the overall improvement of the patient's cardiac function, avoiding the limitations of single-indicator assessment. Simultaneously, by adjusting the weight coefficients of different indicators, personalized assessments can be conducted for different types of heart diseases, improving the accuracy and specificity of the assessment. Furthermore, the quantitative scoring results facilitate intuitive understanding of rehabilitation progress for both doctors and patients, contributing to the development of more precise rehabilitation plans.

[0079] In some of the schemes described above in this application, predictive physiological data representing the target patient under multi-drug co-intervention is generated using a multi-drug effect-cardiac response coupled network model. This model includes: a drug node layer, used to represent each drug in the multi-drug treatment data as an independent node, with each node storing the pharmacokinetic and pharmacodynamic parameters of the drug; an interaction edge layer, used to establish connections between drug nodes with drug interactions, where the weight of the connection edge represents the strength and direction of the drug-drug interaction; a target fusion layer, used to nonlinearly integrate the pharmacodynamic values ​​of multiple drugs acting on the same cardiac receptor target according to their respective target affinity and occupancy, and calculate the overall activation or inhibition degree of the target; and a physiological response layer, used to calculate and output predictive physiological data based on the overall activation or inhibition degree of each target, combined with baseline physiological index data and individualized parameter data.

[0080] Specifically, the drug node layer stores the kinetic parameters of each drug independently, preserving drug-specific information; the interaction edge layer dynamically represents the inhibitory or synergistic relationships between drugs through weighted connections; the target fusion layer integrates the effects of multiple drugs on the same target using nonlinear functions, such as calculating the receptor occupancy superposition effect using the Hill equation; and the physiological response layer couples the integrated target activation level with the patient's baseline physiological indicators, ultimately outputting predicted physiological data that includes the synergistic effects of multiple drugs. This model effectively eliminates prediction bias in multi-drug treatment scenarios by processing drug interactions and target cascade effects in a hierarchical manner, making the generated predicted physiological data closer to the patient's actual physiological state.

[0081] As a preferred embodiment, the solution of this application is specifically implemented as follows:

[0082] When generating predictive physiological data representing target patients under multi-drug co-intervention, a multi-drug effect-cardiac response coupled network model is used. This model includes a drug node layer, an interaction edge layer, a target fusion layer, and a physiological response layer.

[0083] The drug node layer represents each drug in the multi-drug therapy data as an independent node. For example, for a patient using a beta-blocker, a calcium channel blocker, and a diuretic, nodes for each of these three drugs are created in the drug node layer. Each node stores the drug's pharmacokinetic and pharmacodynamic parameters, such as the drug's half-life, clearance rate, and maximum effect size.

[0084] The interaction edge layer establishes connections between drug nodes that exhibit drug interactions. The weight of the connection edge represents the strength and direction of the interaction between the drugs. For example, a synergistic antihypertensive effect may exist between a β-blocker and a calcium channel blocker, so a positive connection edge is established between these two nodes.

[0085] The target fusion layer nonlinearly integrates the pharmacodynamic values ​​of multiple drugs acting on the same cardiac receptor target according to their respective target affinity and occupancy, calculating the overall activation or inhibition degree of the target. For example, for two drugs acting on the β receptor simultaneously, the overall inhibition degree of the β receptor is calculated based on their receptor affinity and concentration.

[0086] The physiological response layer calculates and outputs predicted physiological data based on the overall activation or inhibition level of each target, combined with baseline physiological index data and individualized parameter data. For example, based on the degree of inhibition of β receptors and calcium channels, combined with the patient's baseline heart rate and individualized parameters, it predicts the patient's heart rate changes under the influence of multiple drugs.

[0087] Through the aforementioned technical solution, this application can accurately simulate the complex interactions of multiple drugs in a patient's body and predict their combined impact on cardiac function. This method considers drug interactions, target-level integrative effects, and individualized factors, improving the accuracy of cardiac function prediction in patients undergoing multidrug therapy. This provides a reliable foundation for subsequent removal of drug interference and assessment of actual rehabilitation effects, contributing to a more precise evaluation of cardiac rehabilitation outcomes and guiding the development of personalized rehabilitation programs.

[0088] In summary, the cardiac rehabilitation effect assessment method provided in this application solves the problem of assessment distortion caused by drug interference in traditional assessment methods by constructing a full-chain technical solution of "multi-drug effect modeling - physiological response prediction - drug interference decoupling - real effect assessment".

[0089] Exemplary System

[0090] Figure 2The illustration shows a cardiac rehabilitation effect evaluation system according to an embodiment of this application, including: an acquisition module for acquiring physiological monitoring data, individualized parameter data, comorbidity data, multidrug therapy data, and pharmacokinetic parameters and drug-efficacy interaction parameters of the target patient; an individual indicator generation module for generating baseline physiological indicator data representing the target patient in the absence of drug intervention based on the individualized parameter data and comorbidity data, using a preset individualized cardiac response baseline model; and a first calculation module for calculating the effective concentration of each drug in the target patient based on the multidrug therapy data and pharmacokinetic parameters, using a preset pharmacokinetic model; and a second... The two modules are: a calculation module for calculating the comprehensive pharmacodynamic data under the combined action of multiple drugs based on the concentration and pharmacodynamic interaction parameters using a pre-set pharmacodynamic model; a prediction module for generating predicted physiological data representing the target patient under the combined intervention of multiple drugs based on the comprehensive pharmacodynamic data and baseline physiological index data; an interference separation module for obtaining pure physiological data after removing drug interference based on the predicted physiological data and physiological monitoring data through adaptive filtering and frequency domain decomposition; and an evaluation module for generating a rehabilitation effect score representing the actual improvement of cardiac function in the target patient after removing drug interference based on the pure physiological data using a pre-trained long short-term memory network model.

[0091] In one example, the effective concentration of each drug in the target patient is calculated using a pre-defined pharmacokinetic model. Specifically, this includes: extracting competitive inhibitory relationships between drugs based on multidrug therapy data, which are generated through cytochrome enzyme systems, plasma protein binding sites, and renal clearance pathways; calculating the net clearance rate of each drug in the case of multidrug coexistence based on the competitive inhibitory relationships and pharmacokinetic parameters; and calculating the effective concentration of each drug in the target patient over time based on the net clearance rate.

[0092] In one example, the comprehensive pharmacodynamic data under the combined action of multiple drugs is calculated using a pre-set pharmacodynamic model. Specifically, this includes: calculating the independent pharmacodynamic values ​​of each drug on cardiac function based on the effective concentration of each drug, as well as the pre-set maximum pharmacodynamic parameters and pre-set half-efficacy concentration parameters of each drug; determining the interaction relationship between any two drugs based on pharmacodynamic interaction parameters, including synergistic, additive, and antagonistic effects; calculating the combined pharmacodynamic value of the interactions between the drugs based on the interaction relationship and the independent pharmacodynamic values; and integrating the pharmacodynamic values ​​after integrating all drug interactions to obtain the comprehensive pharmacodynamic data under the combined action of multiple drugs.

[0093] In one example, pure physiological data after removing drug interference is obtained through adaptive filtering and frequency domain decomposition. Specifically, this includes: performing state estimation based on predicted physiological data and physiological monitoring data using an adaptive Kalman filter algorithm to remove observation noise and generate filtered physiological data; performing wavelet transform on the filtered physiological data to extract signal components within a preset frequency bandwidth corresponding to the long-term trend of cardiac rehabilitation improvement as pure physiological data.

[0094] In one example, a rehabilitation effect score is generated using a pre-trained long short-term memory network model. Specifically, this involves: acquiring the target patient's exercise history data; combining the pure physiological data, exercise history data, and individualized parameter data to form a time-series feature vector; inputting the time-series feature vector into the pre-trained long short-term memory network model, which identifies temporal patterns of rehabilitation improvement and outputs multi-dimensional physiological improvement indicators, including resting heart rate improvement, post-exercise heart rate recovery improvement, heart rate variability improvement, exercise tolerance improvement, and symptom improvement score; and calculating the rehabilitation effect score based on the resting heart rate improvement, post-exercise heart rate recovery improvement, heart rate variability improvement, exercise tolerance improvement, and symptom improvement score using a pre-defined weighted scoring function.

[0095] In one example, the preset weighted scoring function is: ;in, This represents the improvement in resting heart rate. This represents the improvement in heart rate recovery after exercise. This represents the improvement in heart rate variability. For improvement in exercise endurance, Assess symptom improvement scores. , , , and The preset weighting coefficients satisfy... .

[0096] In one example, predictive physiological data representing a target patient under multi-drug co-intervention is generated using a multi-drug effect-cardiac response coupled network model. This model includes: a drug node layer, which represents each drug in the multi-drug treatment data as an independent node, with each node storing the drug's pharmacokinetic and pharmacodynamic parameters; an interaction edge layer, which establishes connections between drug nodes that interact with each other, with the weights of the edges representing the strength and direction of the drug-drug interactions; a target fusion layer, which nonlinearly integrates the pharmacodynamic values ​​of multiple drugs acting on the same cardiac receptor target according to their respective target affinity and occupancy, calculating the overall activation or inhibition level of the target; and a physiological response layer, which calculates and outputs predictive physiological data based on the overall activation or inhibition level of each target, combined with baseline physiological index data and individualized parameter data.

[0097] Exemplary electronic devices

[0098] Figure 3 A block diagram of an electronic device according to an embodiment of this application is illustrated.

[0099] like Figure 3 As shown, the electronic device includes one or more processors and memory.

[0100] A processor can be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and can control other components in an electronic device to perform desired functions.

[0101] The memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0102] In one example, the electronic device may also include input devices and output devices, which are interconnected via a bus system and / or other forms of connection mechanism (not shown).

[0103] Of course, for the sake of simplicity, Figure 3 Only some of the components of the electronic device relevant to this application are shown in this illustration; components such as buses, input / output interfaces, etc., are omitted. In addition, the electronic device may include any other suitable components depending on the specific application.

[0104] Exemplary computer-readable media

[0105] Embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps described in the "Exemplary Methods" section above according to the various embodiments of this application.

[0106] Computer-readable storage media may take the form of any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0107] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0108] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0109] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0110] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0111] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for evaluating the effects of cardiac rehabilitation, characterized by, The method comprises the following steps: acquiring physiological monitoring data, individualized parameter data, comorbidity data, multi-drug treatment data and pharmacokinetic interaction parameters and pharmacodynamic interaction parameters of the used drugs of a target patient; generating baseline physiological index data representing the target patient under no drug intervention based on the individualized parameter data and the comorbidity data through a preset individualized heart response baseline model; calculating the effect concentration of each drug in the target patient's body based on the multi-drug treatment data and the pharmacokinetic interaction parameters through a preset pharmacokinetic model; calculating comprehensive pharmacodynamic data under the combined action of multiple drugs based on the effect concentration and the pharmacodynamic interaction parameters through a preset pharmacodynamic model; generating predicted physiological data of the target patient under the combined intervention of multiple drugs based on the comprehensive pharmacodynamic data and the baseline physiological index data; obtaining pure physiological data after removing drug interference based on the predicted physiological data and the physiological monitoring data through adaptive filtering and frequency domain decomposition processing; generating a rehabilitation effect score representing the real improvement of the target patient's heart function after removing drug interference based on the pure physiological data through a pre-trained long short-term memory network model; wherein the adaptive filtering and frequency domain decomposition processing to obtain the pure physiological data after removing drug interference specifically comprises: state estimation through an adaptive Kalman filtering algorithm based on the predicted physiological data and the physiological monitoring data to remove observation noise and generate filtered physiological data; wavelet transform is performed on the filtered physiological data, and signal components within a preset frequency bandwidth corresponding to long-term trends of heart rehabilitation improvement are extracted as the pure physiological data.

2. The cardiac rehabilitation effect evaluation method according to claim 1, characterized by: The preset pharmacokinetic model is used to calculate the effect concentration of each drug in the target patient's body, which specifically comprises: extracting competitive inhibition relationships between drugs through cytochrome enzyme systems, plasma protein binding sites and renal clearance pathways based on the multi-drug treatment data; calculating the net clearance rate of each drug under the condition of multiple drugs coexisting based on the competitive inhibition relationships and the pharmacokinetic interaction parameters; calculating the effect concentration of each drug in the target patient's body over time based on the net clearance rate.

3. The cardiac rehabilitation effect evaluation method according to claim 1, characterized by, The preset pharmacodynamic model is used to calculate the comprehensive pharmacodynamic data under the combined action of multiple drugs, which specifically comprises: calculating independent pharmacodynamic values of each drug on heart function under individual action based on the effect concentration of each drug and preset maximum pharmacodynamic parameters and preset half-effective concentration parameters of each drug; judging the interaction relationship between any two drugs based on the pharmacodynamic interaction parameters, wherein the interaction relationship includes synergistic effect, additive effect and antagonistic effect; calculating combined pharmacodynamic values of the interaction between two drugs based on the interaction relationship and the independent pharmacodynamic values; integrating the pharmacodynamic values of all drug interactions to obtain the comprehensive pharmacodynamic data under the combined action of multiple drugs.

4. The cardiac rehabilitation effect evaluation method according to claim 1, characterized by, The pre-trained long short-term memory network model is used to generate a rehabilitation effect score, which specifically comprises: acquiring exercise history data of the target patient; combine the pure physiological data, the exercise history data and the individualized parameter data to form a time series feature vector; input the time series feature vector into the pre-trained long short-term memory network model, identify a time series pattern of rehabilitation improvement through a long short-term memory network, and output a plurality of dimensions of physiological improvement indexes, the plurality of dimensions of physiological improvement indexes including a resting heart rate improvement value, a post-exercise heart rate recovery improvement value, a heart rate variability improvement value, an exercise tolerance improvement value and a symptom improvement score; calculate the rehabilitation effect score through a preset weighted score function based on the resting heart rate improvement value, the post-exercise heart rate recovery improvement value, the heart rate variability improvement value, the exercise tolerance improvement value and the symptom improvement score.

5. The cardiac rehabilitation effect evaluation method according to claim 4, characterized by, The preset weighted score function is: ; wherein, is the rest heart rate improvement value, is the post-exercise heart rate recovery improvement value, is the heart rate variability improvement value, is the exercise tolerance improvement value, is the symptom improvement score, , , , and is a preset weight coefficient, satisfying .

6. The cardiac rehabilitation effect evaluation method according to claim 1, characterized by, The generation of the predicted physiological data of the target patient under the joint intervention of multiple drugs adopts a multi-drug effect-heart response coupling network model, which includes: a drug node layer for representing each drug in the multiple-drug treatment data as an independent node, each node storing pharmacokinetic action parameters and pharmacodynamic action parameters of the drug; an interaction edge layer for establishing a connection edge between drug nodes in the presence of drug interactions, the weight of the connection edge representing the strength and direction of the interaction between drugs; a target point fusion layer for, for multiple drugs acting on the same cardiac receptor target point, nonlinearly integrating the drug efficacy values according to the respective target point affinities and occupancy rates to calculate the comprehensive activation or inhibition degree of the target point; a physiological response layer for calculating and outputting the predicted physiological data according to the comprehensive activation or inhibition degree of each target point, combining the baseline physiological indicator data and the individualized parameter data.

7. A cardiac rehabilitation effectiveness evaluation system characterized by, It includes: an acquisition module for acquiring physiological monitoring data, individualized parameter data, comorbidity data, multiple-drug treatment data and pharmacokinetic action parameters and pharmacodynamic interaction parameters between drugs used; an individual indicator generation module for generating baseline physiological indicator data representing the target patient under no drug intervention according to the individualized parameter data and the comorbidity data through a preset individualized cardiac response baseline model; a first calculation module for calculating the action concentration of each drug in the target patient's body according to the multiple-drug treatment data and the pharmacokinetic action parameters through a preset pharmacokinetic model; a second calculation module for calculating comprehensive pharmacodynamic data under the joint action of multiple drugs according to the action concentration and the pharmacodynamic interaction parameters through a preset pharmacodynamic model; a prediction module for generating predicted physiological data representing the target patient under the joint intervention of multiple drugs according to the comprehensive pharmacodynamic data and the baseline physiological indicator data; an interference separation module for obtaining pure physiological data removed of drug interference based on the predicted physiological data and the physiological monitoring data through adaptive filtering and frequency domain decomposition processing; an evaluation module for generating a rehabilitation effect score representing the real improvement of the target patient's cardiac function after removing drug interference through a pre-trained long short-term memory network model based on the pure physiological data. The pure physiological data after removing the drug interference is obtained through adaptive filtering and frequency domain decomposition processing, and specifically includes: Based on the predicted physiological data and the physiological monitoring data, state estimation is performed through an adaptive Kalman filtering algorithm to remove observation noise and generate filtered physiological data; Wavelet transform is performed on the filtered physiological data, and a signal component within a preset frequency bandwidth corresponding to a long-term trend of cardiac rehabilitation improvement is extracted as the pure physiological data. 8.An electronic device comprising a memory and a processor, the electronic device characterized by: The memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions, and the computer executable instructions, when executed by the processor, implement the steps of the method in any one of claims 1-6.

9. A computer storage medium having stored thereon computer- executable instructions, comprising: The computer executable instructions, when executed by the processor, implement the steps of the method in any one of claims 1-6.

Images