Monitoring management method and system for medical health

Abstract

The invention discloses a monitoring management method and system for medical health, and relates to the technical field of medical health, and the method comprises the steps: receiving an original physiological signal from a wearable device, defining a dynamic adaptive filter network, and generating purified physiological data based on the dynamic adaptive filter network in combination with the original physiological signal; and enhancing the relevance between the purified physiological data and the user behavior data by using a multi-modal fusion attention mechanism and a generative adversarial network, and determining a multi-modal health perception mechanism. According to the invention, the influence of environmental noise on physiological signals is effectively reduced through the dynamic adaptive filter network, and the monitoring precision is improved. Through the multi-modal fusion attention mechanism and the generative adversarial network, the relevance between the physiological data and the user behavior data is enhanced, so that the system can more comprehensively understand the health state of the user.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical health, in particular to a monitoring management method and system for medical health. BACKGROUND

[0002] In the field of medical health monitoring management, with the continuous improvement of people's demand for health management, wearable devices, mobile applications and cloud platform systems have gradually become mainstream technical means and are widely used in real-time monitoring of users' physiological parameters and management and analysis of health data such as heart rate, blood pressure, blood sugar, etc. These technologies can provide health warnings or personalized recommendations by collecting and processing users' health data, and play an important role in disease prevention and health management.

[0003] However, in actual application, the collection of sensor data in the prior art is easily affected by environmental noise, such as motion artifacts and electromagnetic interference, etc., resulting in a decrease in monitoring accuracy. Especially in dynamic scenarios, such as when the user is moving, the error rate of sensor data is as high as 10-20%, which seriously affects the reliability of the diagnosis result and the effectiveness of health management. Secondly, the data processing architecture of the existing system mostly adopts a centralized cloud analysis mode, which has strong computing power, but the delay problem of data transmission and processing is prominent, and the response time is usually more than 500ms. This delay can lead to a lack of timely intervention in emergency situations such as abnormal heart rate and sudden illness, thereby endangering the user's life safety. In addition, the existing technology also has deficiencies in real-time performance and adaptive ability in complex scenarios, and it is difficult to meet the diversified needs of health management. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a monitoring management method and system for medical health to solve the problems of sensor data being easily disturbed by environmental noise, real-time response delay, etc. in the prior art.

[0005] To achieve the above object, the present application is implemented by the following technical solutions: In a first aspect, the present application provides a monitoring and management method for medical health, comprising: receiving original physiological signals from a wearable device, and defining a dynamic adaptive filtering network; generating purified physiological data based on the dynamic adaptive filtering network in combination with the original physiological signals; utilizing a multi-modal fusion attention mechanism and a generative adversarial network to enhance the relevance between the purified physiological data and user behavior data, and determining a multi-modal health perception mechanism; based on the multi-modal health perception mechanism, constructing a distributed edge computing feedback system, processing different levels of feedback signals using the distributed edge computing feedback system, and generating an optimized health state prediction path; based on the distributed edge computing feedback system, constructing a task adaptive module using a meta-learning framework and a deep reinforcement learning algorithm, adjusting key parameters in the task adaptive module based on the task adaptive module and the optimized health state prediction path, and generating a dynamic sampling frequency adjustment strategy; based on the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, constructing a personalized health management module using an incremental context modeling technology based on a spatio-temporal graph neural network, and constructing a medical health monitoring and management system based on the personalized health management module, the dynamic adaptive filtering network, the multi-modal health perception mechanism, the distributed edge computing feedback system, and the task adaptive module.

[0006] Optionally, based on the distributed edge computing feedback system, a task adaptive module is constructed using a meta-learning framework and a deep reinforcement learning algorithm, based on the task adaptive module and the optimized health state prediction path, key parameters in the task adaptive module are adjusted to generate a dynamic sampling frequency adjustment strategy, which comprises: constructing an initial task adaptive module based on the distributed edge computing feedback system using a meta-learning framework and a deep reinforcement learning algorithm, and optimizing the switching efficiency of the initial task adaptive module between different tasks through a Q-learning algorithm and a policy gradient method; based on the optimized health state prediction path, a variational autoencoder is used to analyze the change trend from the data distribution of different levels of feedback signals in the distributed edge computing feedback system, and the key parameters in the initial task adaptive module are dynamically adjusted through Bayesian optimization to generate a target task adaptive module; the target task adaptive module adjusts the sampling strategy according to the health background information of different users using a context perception mechanism to generate a dynamic sampling frequency adjustment strategy.

[0007] Optionally, the context-aware mechanism is used to adjust the sampling strategy of the target task adaptive module according to the health background information of different users, and a dynamic sampling frequency adjustment strategy is generated, including: using the context-aware mechanism, combining a spatio-temporal graph neural network and an attention mechanism, and analyzing key features from the health background information of different users in the distributed edge computing feedback system, the key features including user age, medical history records, and physiological index fluctuation range; based on the key features, combining Bayesian optimization and Gaussian process regression, dynamically adjusting the sampling frequency in the target task adaptive module to generate a preliminary dynamic sampling frequency adjustment strategy; using a time series prediction model, optimizing the preliminary dynamic sampling frequency adjustment strategy through deep reinforcement learning to generate an optimized dynamic sampling frequency adjustment strategy; combining a genetic algorithm and a differential evolution algorithm, searching for an optimal sampling frequency adjustment strategy from the optimized dynamic sampling frequency adjustment strategy; introducing an active learning mechanism to select target data from the optimal sampling frequency adjustment strategy to generate a dynamic sampling frequency adjustment strategy.

[0008] Optionally, the context-aware mechanism is used to analyze key features from the health background information of different users in the distributed edge computing feedback system, combining a spatio-temporal graph neural network and an attention mechanism, including: using data mining technology to collect and preprocess the health background information of different users in the distributed edge computing feedback system to obtain user health background data of the distributed edge computing feedback system; introducing a context-aware mechanism, combining principal component analysis and factor analysis, and performing preliminary screening and dimensionality reduction processing on the user health background data to obtain target user health background information; applying a spatio-temporal graph neural network and complex network analysis to analyze the health relationships between users from the target user health background information to construct a user health relationship graph; applying an attention mechanism, combining deep reinforcement learning, to highlight important features in the user health relationship graph to obtain key feature weight allocation results; analyzing the user health relationship graph and the key feature weight allocation results to extract key features.

[0009] Optionally, the predicting the path based on the optimized health state, using a variational autoencoder, analyzing the trend of change from the data distribution of the feedback signals of different levels in the distributed edge computing feedback system, and dynamically adjusting the key parameters in the initial task adaptive module through Bayesian optimization to generate a target task adaptive module, comprising: according to the optimized health state prediction path, using a variational autoencoder and a statistical analysis tool, modeling the data distribution of feedback signals of different levels in the distributed edge computing feedback system, and analyzing potential patterns and change rules from the feedback information to generate a data distribution change trend; based on the data distribution change trend, dynamically adjusting the key parameters in the initial task adaptive module through Bayesian optimization, Gaussian process regression and gradient boosting tree to generate a target task adaptive module.

[0010] Optionally, the use of a multi-modal fusion attention mechanism and a generative adversarial network to enhance the correlation between the purified physiological data and user behavior data, and determine a multi-modal health perception mechanism, comprising: using a multi-level spatio-temporal graph neural network to define an initial dynamic adaptive filtering network, and analyzing an initial health association from the internal relationship between the physiological data and user behavior data of the initial dynamic adaptive filtering network; based on the initial health association, introducing a multi-modal fusion attention mechanism, combining a bidirectional long short-term memory network to jointly encode and process the physiological data and corresponding behavior data in the initial health association to obtain a multi-modal health information representation; according to the multi-modal health information representation, constructing a generative adversarial network, and analyzing key data from the adversarial training process between the generator and the discriminator in the generative adversarial network; based on the key data, combining a variational autoencoder to optimize the initial dynamic adaptive filtering network to obtain a target dynamic adaptive filtering network, and determining a multi-modal health perception mechanism based on the target dynamic adaptive filtering network.

[0011] Optionally, based on the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, a personalized health management module is constructed using incremental context modeling technology based on a spatio-temporal graph neural network, including: using the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, combining time series analysis and Bayesian optimization to determine an incremental update rule; based on the incremental update rule, using incremental context modeling technology based on a spatio-temporal graph neural network, combining a hierarchical clustering algorithm, modeling and processing multi-modal data in the multi-modal health perception mechanism to obtain a multi-modal data structured representation; based on the multi-modal data structured representation, combining a dynamic graph convolution network and a variational autoencoder, encoding data obtained from the distributed edge computing feedback system and inputting the data into the multi-modal data structured representation to output an updated context representation; applying an attention mechanism, combining deep reinforcement learning to highlight key features in the updated context representation to obtain an optimized context representation; combining reinforcement learning and genetic algorithm to optimize and adjust parameters in the incremental update process in the task adaptive module to generate optimized parameter configurations; and based on the optimized context representation and the optimized parameter configurations, generating a personalized health management module.

[0012] In a second aspect, the present application provides a monitoring and management system for medical health, comprising: a receiving module for receiving raw physiological signals from a wearable device and defining a dynamic adaptive filtering network, based on which the raw physiological signals are combined to generate purified physiological data; an enhancement module for using a multi-modal fusion attention mechanism and a generative adversarial network to enhance the correlation between the purified physiological data and user behavior data, and determining a multi-modal health perception mechanism; a construction module for constructing a distributed edge computing feedback system based on the multi-modal health perception mechanism, using the distributed edge computing feedback system to process different levels of feedback signals to generate an optimized health state prediction path; an adjustment module for using a meta-learning framework and a deep reinforcement learning algorithm to construct a task adaptive module based on the distributed edge computing feedback system and the optimized health state prediction path, adjusting key parameters in the task adaptive module based on the task adaptive module and the optimized health state prediction path to generate a dynamic sampling frequency adjustment strategy; and a utilization module for using incremental context modeling technology based on a spatio-temporal graph neural network to construct a personalized health management module based on the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, and constructing a medical health monitoring and management system based on the personalized health management module, the dynamic adaptive filtering network, the multi-modal health perception mechanism, the distributed edge computing feedback system, and the task adaptive module.

[0013] In a third aspect, the present application provides a computing device comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the computer program to perform the method for monitoring and management of medical health according to any one of the first aspect.

[0014] In a fourth aspect, the present application provides a computer storage medium storing computer program instructions, wherein the computer program instructions are executed by a processor to implement the method for monitoring and management of medical health according to any one of the first aspect. Advantages

[0015] The present application effectively reduces the influence of environmental noise on physiological signals through a dynamic adaptive filtering network, improving monitoring accuracy. Through a multi-modal fusion attention mechanism and a generative adversarial network, the correlation between physiological data and user behavior data is enhanced, enabling the system to more comprehensively understand the user's health status. The distributed edge computing feedback system enables real-time response, meeting the immediate intervention needs in emergency situations. The task adaptive module, combined with the optimized health status prediction path, dynamically adjusts the sampling frequency, reducing energy consumption and improving system flexibility. Based on the personalized health management module, the system can gradually introduce new data, maintaining existing performance while continuously improving, ensuring long-term effectiveness and personalized services. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 FIG. 1 is a structural schematic diagram of the monitoring and management system of the present application. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0018] As shown in FIG. 1, the medical health monitoring and management system of the present application comprises a receiving module, an enhancing module, a constructing module, an adjusting module, and a utilizing module. Figure 1

[0019] Firstly, the receiving module receives the original physiological signals from the wearable device and defines a dynamic adaptive filtering network to generate purified physiological data. The core of the dynamic adaptive filtering network lies in its dynamic weight adjustment mechanism, which can optimize the filtering effect in real time according to the changes of environmental noise.

[0020] Specifically, the dynamic adaptive filtering formula is: wherein ​Indicates physiological signals after purification. Represents primitive physiological signals, Indicates dynamic weights. This indicates the bias term.

[0021] In practical applications, the receiving module detects noise frequency and amplitude fluctuations in the raw physiological signal in real time, such as 10-50Hz high-frequency noise from motion artifacts and 50Hz power frequency noise from electromagnetic interference. It then analyzes the time-frequency distribution characteristics of the noise using Short Time Fourier Transform (STFT) and dynamically adjusts the weight parameters of the filtering network accordingly. and bias terms For example, when motion is detected, the filter intensity in the frequency band above 20Hz is automatically increased to quickly filter motion noise, thereby effectively reducing the impact of environmental noise on signal quality. For example, during user movement, the dynamic adaptive filtering network can quickly identify and filter out high-frequency noise caused by motion, ensuring the accuracy of the output signal.

[0022] Next, the enhancement module utilizes a multimodal fusion attention mechanism and a generative adversarial network to strengthen the correlation between purified physiological data and user behavior data, thus establishing a multimodal health perception mechanism. The core of the multimodal fusion attention mechanism lies in its ability to highlight the importance of key features through joint encoding of data from different modalities, thereby improving the system's perception capabilities.

[0023] Specifically, the formula for the multimodal fusion attention mechanism is: and ,in Indicates attention weights, This represents the unnormalized attention score. and These represent the eigenvectors of the i-th and j-th modes, respectively. and This represents the learnable parameters.

[0024] In practical applications, the enhancement module first defines an initial dynamic adaptive filtering network using a multi-level spatiotemporal graph neural network to extract initial health associations from the intrinsic connections between physiological data and user behavior data. Subsequently, a multimodal fusion attention mechanism is introduced, combined with a bidirectional long short-term memory network, to jointly encode the physiological data and corresponding behavioral data in the initial health associations, resulting in a multimodal health information representation. Based on this, a generative adversarial network is constructed, and the data quality is further optimized through an adversarial training process between the generator and discriminator, ultimately determining the target dynamic adaptive filtering network and the multimodal health perception mechanism.

[0025] The construction module is based on a multi-modal health perception mechanism to construct a distributed edge computing feedback system for processing feedback signals from different levels to generate an optimized health state prediction path. The design goal of the distributed edge computing feedback system is to achieve low latency and high reliability in health state prediction. In practical applications, the construction module first processes the feedback signals in the distributed edge computing feedback system in layers, dividing the signals into three categories according to priority: real-time signals, historical signals, and background signals. Specifically, the distributed edge computing feedback system processes signals according to response priority: real-time signals: emergency data that needs to be processed within milliseconds, such as heart rate > 180 beats / min and blood oxygen < 90%; historical signals: periodic data for trend analysis, such as average blood glucose in the past 7 days and sleep cycle; background signals: static data that assist in context, such as user gender and underlying disease type. Real-time signals are processed locally by edge nodes with a delay < 100ms, while historical and background signals are stored and analyzed in the cloud to achieve layered response.

[0026] Subsequently, time series analysis and Bayesian optimization techniques are used to model and predict various signals to generate an optimized health state prediction path. For example, in the cardiovascular disease monitoring scenario, the distributed edge computing feedback system can quickly predict potential health risks by analyzing real-time heart rate variability and blood pressure fluctuations, and promptly alert users.

[0027] The adjustment module is based on the distributed edge computing feedback system and uses a meta-learning framework and deep reinforcement learning algorithm to construct a task-adaptive module. It adjusts key parameters in the task-adaptive module based on the optimized health state prediction path to generate a dynamic sampling frequency adjustment strategy. The core of the task-adaptive module is its ability to dynamically adjust the sampling frequency based on different users' health background information, thereby reducing energy consumption while ensuring monitoring accuracy.

[0028] In practical applications, the adjustment module first constructs an initial task-adaptive module using a meta-learning framework and deep reinforcement learning algorithm, and optimizes the switching efficiency of the initial task-adaptive module between different tasks using Q-learning algorithm and policy gradient method. Specifically, in the policy gradient method, the update formula of the policy network parameter is as follows: where is the trajectory (state-action sequence), is the action probability output by the policy network, is the action value function. Subsequently, based on the optimized health state prediction path, the data distribution of the feedback signals from different levels of the distributed edge computing feedback system is analyzed to find the trend of change using the variational autoencoder, and the key parameters in the initial task adaptive module are dynamically adjusted through Bayesian optimization to generate the target task adaptive module. For example, in the blood glucose monitoring scenario of diabetic patients, the task adaptive module can dynamically adjust the sampling frequency of blood glucose monitoring according to the patient's dietary habits, exercise frequency, and other health background information, thereby extending the device's battery life while meeting monitoring needs.

[0029] Using the module, based on the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, the incremental context modeling technology based on spatio-temporal graph neural networks is used to construct a personalized health management module, and combined with the dynamic adaptive filtering network, the multi-modal health perception mechanism, the distributed edge computing feedback system, and the task adaptive module, a complete medical health monitoring and management system is constructed. The design goal of the personalized health management module is to realize long-term tracking and dynamic optimization of the user's health status.

[0030] In practical applications, the module first uses the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, combined with time series analysis and Bayesian optimization to determine the incremental update rule. Subsequently, based on the incremental update rule, the incremental context modeling technology based on spatio-temporal graph neural networks is used, combined with hierarchical clustering algorithm to model and process the multi-modal data in the multi-modal health perception mechanism, to obtain the structured representation of multi-modal data. On this basis, combined with dynamic graph convolution network and variational autoencoder, the data obtained from the distributed edge computing feedback system is encoded, and the data is input into the structured representation of multi-modal data to output the updated context representation. Finally, the attention mechanism is applied to highlight the key features in the updated context representation using deep reinforcement learning, to obtain the optimized context representation, and the parameters in the incremental update process in the task adaptive module are optimized and adjusted using reinforcement learning and genetic algorithm to generate the optimized parameter configuration. For example, in the chronic disease management scenario, the personalized health management module can dynamically adjust the intervention strategy according to the user's long-term health data, thereby realizing precise and personalized health management services.

[0031] In summary, the present application realizes the complete process from original physiological signal collection to personalized health management through the receiving module, the enhancement module, the construction module, the adjustment module and the utilization module. In actual application scenarios, the medical health monitoring management system of the present application can significantly improve the monitoring accuracy, reduce the real-time response delay, and has the ability to efficiently process complex health tasks. For example, in the remote medical scenario, the present system can realize real-time monitoring of the patient's health status through the distributed edge computing feedback system, and provide accurate diagnosis and treatment suggestions for doctors through the personalized health management module. In addition, the present application also effectively solves the problems of existing technologies, such as sensor data being easily disturbed by environmental noise, real-time response delay, etc., through the dynamic self-adaptive filtering network, the multi-modal fusion attention mechanism, and the generative adversarial network.

[0032] It should be noted that, in this paper, relational terms such as first and second are used merely to distinguish one entity or action from another, without necessarily requiring or implying that there is any such actual relationship or order between or among the entities or actions. Also, the terms "comprises", "comprising", or any other variations thereof are intended to cover non-exclusive inclusions, so that a process, method, article, or apparatus that includes a list of elements does not only include those elements, but also includes other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0033] Although embodiments of the present application have been shown and described, it will be understood by those having ordinary skill in the art that various changes, modifications, alternatives, and variations can be made thereto without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring management of medical health, characterized by, The method comprises the following steps: receiving raw physiological signals from a wearable device and defining a dynamic adaptive filtering network, generating purified physiological data based on the dynamic adaptive filtering network combined with the raw physiological signals; using a multi-modal fusion attention mechanism and a generative adversarial network to enhance the correlation between the purified physiological data and user behavior data, and determining a multi-modal health perception mechanism; based on the multi-modal health perception mechanism, constructing a distributed edge computing feedback system, processing different levels of feedback signals from the distributed edge computing feedback system to generate an optimized health state prediction path; based on the distributed edge computing feedback system, using a meta-learning framework and a deep reinforcement learning algorithm to construct a task adaptive module, adjusting the key parameters in the task adaptive module based on the task adaptive module and the optimized health state prediction path to generate a dynamic sampling frequency adjustment strategy; based on the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, using an incremental context modeling technology based on a spatio-temporal graph neural network to construct a personalized health management module, and based on the personalized health management module, the dynamic adaptive filtering network, the multi-modal health perception mechanism, the distributed edge computing feedback system and the task adaptive module, constructing a medical health monitoring and management system.

2. The method of claim 1, wherein, The method comprises the following steps: using a meta-learning framework and a deep reinforcement learning algorithm to construct an initial task adaptive module based on the distributed edge computing feedback system, and optimizing the switching efficiency of the initial task adaptive module between different tasks through a Q-learning algorithm and a policy gradient method; based on the optimized health state prediction path, using a variational autoencoder to analyze the trend of data distribution from different levels of feedback signals in the distributed edge computing feedback system, and dynamically adjusting the key parameters in the initial task adaptive module through Bayesian optimization to generate a target task adaptive module; using a context perception mechanism to adjust the sampling strategy of the target task adaptive module according to the health background information of different users to generate a dynamic sampling frequency adjustment strategy.

3. The method of claim 2, wherein, The method comprises the following steps: using a context perception mechanism to analyze key features from the health background information of different users in the distributed edge computing feedback system based on a spatio-temporal graph neural network and an attention mechanism, the key features including user age, medical history records and physiological index fluctuation range; based on the key features, combining Bayesian optimization and Gaussian process regression to dynamically adjust the sampling frequency in the target task adaptive module to generate a preliminary dynamic sampling frequency adjustment strategy; The time series prediction model is used to optimize the preliminary dynamic sampling frequency adjustment strategy through deep reinforcement learning to generate an optimized dynamic sampling frequency adjustment strategy. The genetic algorithm and the differential evolution algorithm are combined to search for an optimal sampling frequency adjustment strategy from the optimized dynamic sampling frequency adjustment strategy. An active learning mechanism is introduced to select target data from the optimal sampling frequency adjustment strategy to generate a dynamic sampling frequency adjustment strategy.

4. The method of claim 3, wherein, The context awareness mechanism is used to analyze key features from the health background information of different users in the distributed edge computing feedback system, including: Data mining techniques are used to collect and preprocess the health background information of different users in the distributed edge computing feedback system to obtain user health background data of the distributed edge computing feedback system. A context awareness mechanism is introduced to combine principal component analysis and factor analysis to preliminarily screen and reduce the dimensionality of the user health background data to obtain target user health background information. A spatio-temporal graph neural network and complex network analysis are applied to analyze the health relationships between users from the target user health background information to construct a user health relationship graph. An attention mechanism is applied to combine deep reinforcement learning to highlight important features in the user health relationship graph to obtain key feature weight distribution results. The user health relationship graph and the key feature weight distribution results are analyzed to extract key features.

5. The method of claim 2, wherein, Based on the optimized health state prediction path, a variational autoencoder is used to analyze the data distribution of different levels of feedback signals from the distributed edge computing feedback system to generate a target task adaptive module, including: Based on the optimized health state prediction path, a variational autoencoder and statistical analysis tools are used to model the data distribution of different levels of feedback signals from the distributed edge computing feedback system, and potential patterns and change laws are analyzed from the feedback information to generate a data distribution change trend. Based on the data distribution change trend, the key parameters in the initial task adaptive module are dynamically adjusted through Bayesian optimization, Gaussian process regression, and gradient boosting trees to generate a target task adaptive module.

6. The method of claim 1, wherein, The multi-modal fusion attention mechanism and the generative adversarial network are used to enhance the association between the purified physiological data and user behavior data to determine a multi-modal health perception mechanism, including: A multi-level spatio-temporal graph neural network is used to define an initial dynamic adaptive filtering network, and the internal relationship between the physiological data and user behavior data from the initial dynamic adaptive filtering network is analyzed to obtain an initial health association. Based on the initial health association, a multi-modal fusion attention mechanism is introduced to combine a bidirectional long short-term memory network to jointly encode the physiological data and corresponding behavior data in the initial health association to obtain a multi-modal health information representation. According to the multi-modal health information representation, a generative adversarial network is constructed, and key data is analyzed from an adversarial training process between a generator and a discriminator in the generative adversarial network; Based on the key data, a variational autoencoder is combined to optimize the initial dynamic adaptive filtering network to obtain a target dynamic adaptive filtering network, and based on the target dynamic adaptive filtering network, a multi-modal health perception mechanism is determined.

7. The method of claim 1, wherein, Based on the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, a personalized health management module is constructed by using an incremental context modeling technology based on a spatio-temporal graph neural network, including: Based on the optimized health state prediction path and the dynamic sampling frequency adjustment strategy, a time series analysis and Bayesian optimization are combined to determine an incremental update rule; Based on the incremental update rule, an incremental context modeling technology based on a spatio-temporal graph neural network is used in combination with a hierarchical clustering algorithm to model and process multi-modal data in the multi-modal health perception mechanism to obtain a multi-modal data structured representation; According to the multi-modal data structured representation, a dynamic graph convolutional network and a variational autoencoder are combined to encode data obtained from the distributed edge computing feedback system and input the data into the multi-modal data structured representation to output an updated context representation; An attention mechanism is applied in combination with deep reinforcement learning to highlight key features in the updated context representation to obtain an optimized context representation; In combination with reinforcement learning and genetic algorithms, parameters in the incremental update process in the task adaptive module are optimized and adjusted to generate optimized parameter configurations; Based on the optimized context representation and the optimized parameter configurations, a personalized health management module is generated.

8. A monitoring management system for medical health, characterized by, Including: A receiving module for receiving raw physiological signals from a wearable device and defining a dynamic adaptive filtering network, based on which the dynamic adaptive filtering network is combined with the raw physiological signals to generate purified physiological data; An enhancement module for using a multi-modal fusion attention mechanism and a generative adversarial network to enhance the relevance between purified physiological data and user behavior data to determine a multi-modal health perception mechanism; A construction module for constructing a distributed edge computing feedback system based on the multi-modal health perception mechanism, using the distributed edge computing feedback system to process different levels of feedback signals to generate an optimized health state prediction path; An adjustment module for using a meta-learning framework and a deep reinforcement learning algorithm to construct a task adaptive module based on the distributed edge computing feedback system and the optimized health state prediction path, adjusting key parameters in the task adaptive module to generate a dynamic sampling frequency adjustment strategy. The utilization module is used for predicting a path based on the optimized health state and the dynamic sampling frequency adjustment strategy, constructing a personalized health management module by using an incremental context modeling technology based on a space-time graph neural network, and constructing a medical health monitoring management system based on the personalized health management module, the dynamic adaptive filtering network, the multi-modal health sensing mechanism, the distributed edge computing feedback system, and the task adaptive module.

9. A computing device, comprising: The system comprises a processor and a memory, and the memory stores a computer program, and the processor is configured to run the computer program to perform the method for monitoring and managing medical health according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program product, comprising a computer program instruction stored thereon, wherein the computer program instruction is executed by a processor to implement the method for monitoring and managing medical health according to any one of claims 1 to 7.

Images