Home nursing management platform and method for chronic disease patients

Abstract

The invention relates to the technical field of medical health, in particular to a chronic disease patient-oriented home nursing management platform, comprising: a terminal sensing module comprising a monitoring bracelet and a glucometer and used for obtaining real-time data of a patient; the monitoring bracelet is used for monitoring physiological indexes and motion data of a patient in real time; the glucometer is used for monitoring blood glucose data of a patient in real time; the edge calculation module comprises home gateway equipment and an edge intelligent analysis system; the cloud service module comprises a data processing center and an intelligent decision-making system; the intelligent decision-making system comprises a health risk assessment model, an exercise scheme optimization model and an expert knowledge base comprising an exercise guidance strategy and a life rhythm management strategy; and the optimization module is used for comprehensively coordinating the generated results, dynamically adjusting motion scheme parameters according to patient feedback and execution conditions, and pushing an optimized personalized scheme to a user terminal for execution.

Description

Technical Field

[0001] The present invention relates to the technical field of medical health, and specifically to a home care management platform and method for chronic disease patients. Background Art

[0002] The prior art document with the publication number CN117854696A proposes a home care management platform and method for chronic disease patients. Through an integrated platform, data sharing and interconnection of multiple systems such as the public health management system, hospital information system, and family doctor signing system are realized. The data in different systems are integrated to avoid information silos, improve the utilization rate of information and the overall work efficiency. By analyzing the basic medical information of patients, their medical risk levels are determined, and corresponding measures are generated to help medical staff better provide personalized medical services and disease prevention. Real-time health data of patients are dynamically obtained, and new management decisions are generated according to data changes, significantly improving the intelligent level of health management. By integrating medical and public health services, a unified basic medical information management system is established to achieve seamless connection between diagnosis and treatment and prevention. Through the integrated platform, the basic medical information is updated in real time, and the information is synchronized to each relevant system, enabling medical staff to make accurate medical decisions based on the latest patient data.

[0003] However, the prior art still has some deficiencies. Most chronic disease patients need appropriate exercise, but the prior art cannot evaluate the health risks of chronic disease patients during home exercise, cannot analyze the exercise status in combination with the disease condition, and cannot analyze the sleep movement status in combination with the disease condition.

[0004] In view of this, we propose a home care management platform and method for chronic disease patients. Summary of the Invention

[0005] The purpose of the present invention is to provide a home care management platform and method for chronic disease patients to solve the problems raised in the above background art.

[0006] To achieve the above purpose, one aspect of the present invention provides the following technical solution:

[0007] The terminal perception module, including a monitoring bracelet and a blood glucose meter, is used to obtain the real-time data of patients. The monitoring bracelet is used to monitor the physiological indicators and exercise data of patients in real time. The monitoring bracelet is integrated with a heart rate sensor, a blood pressure sensor, a blood oxygen sensor, a body movement sensor, and a pedometer for collection. The blood glucose meter is used to monitor the blood glucose data of patients in real time, and the blood glucose meter supports NFC wireless transmission function;

[0008] Edge computing module, including a home gateway device and an edge intelligent analysis system; the home gateway device is used to collect real-time data of the terminal sensing module, and the home gateway device includes a communication module integrating Bluetooth, WiFi, and ZigBee communication protocols;

[0009] Cloud service module, the cloud service module includes a data processing center and an intelligent decision-making system; the data processing center includes a distributed storage system, a data cleaning and standardization processing module, and a multi-source data fusion analysis module based on deep learning; the intelligent decision-making system includes a health risk assessment model, an exercise plan optimization model, and an expert knowledge base including exercise guidance strategies and life rhythm management strategies; the health risk assessment model uses a graph structure analysis model constructed based on a graph neural network to obtain a multi-dimensional health risk score, and calculates a comprehensive risk index based on the scores of each dimension; the exercise plan optimization model constructs and applies an exercise model according to physiological indicators and real-time data, evaluates the current exercise score, and gives an exercise plan in combination with the exercise guidance strategy of the expert knowledge base;

[0010] Optimization module, comprehensively coordinates the generated results, dynamically adjusts the exercise plan parameters according to the patient's feedback and execution situation, and pushes the optimized personalized plan to the user terminal for execution.

[0011] Preferably, the monitoring bracelet further includes a sleep sensor, and the sleep sensor is used to monitor the patient's sleep data in real time;

[0012] The intelligent decision-making system further includes a work and rest advice generation model, and the work and rest advice generation model evaluates the current sleep score according to the disease type and sleep data, and gives a work and rest advice plan in combination with the life rhythm management strategy of the expert knowledge base.

[0013] Preferably, the edge intelligent analysis system is used for real-time data preprocessing, and realizes fast-response real-time monitoring and early warning locally, including:

[0014] Data access layer, used for real-time data reception, including:

[0015] Multi-channel data cache pool, used for parallel reception and real-time data;

[0016] Data frame synchronization unit, used for timestamp alignment of real-time data of different sensors;

[0017] Data quality assessment unit, used for filtering abnormal data frames;

[0018] Real-time computing layer, including:

[0019] The feature extraction processing unit includes: a parallel Fourier transform module for extracting the frequency-domain features of physiological signals; a sliding window statistics module for calculating the time-domain statistical features; a multi-scale wavelet decomposition module for extracting the multi-scale features of signals;

[0020] The fast detection processing unit includes: a signal prediction module based on Kalman filtering; an anomaly detection module based on a lightweight neural network; a threshold determination module based on a rule engine;

[0021] The early warning decision layer adopts a hierarchical processing mechanism, including:

[0022] The lightweight early warning evaluation unit includes: a fast classification module based on a decision tree; a risk scoring module based on fuzzy logic; an early warning level determination module based on a rule library;

[0023] The early warning trigger control unit includes: an early warning priority sorting module; an early warning information merging module; an early warning sending control module.

[0024] Preferably, the distributed storage system adopts a multi-node hierarchical storage architecture, stores real-time data in cache nodes, and distributes historical data to storage nodes with different performance levels according to the time span, realizing efficient access to data;

[0025] The data cleaning and standardization processing module adopts a pipeline processing method, and sequentially performs outlier detection, missing value filling, time alignment, and numerical normalization to ensure data quality and consistency;

[0026] The multi-source data fusion analysis module adopts a deep learning model based on an attention mechanism to perform feature extraction and fusion on real-time data from different sources to obtain corresponding feature vectors.

[0027] Preferably, the health risk assessment model is a multi-layer temporal heterogeneous graph neural network model, including:

[0028] The node feature layer includes: physiological index nodes, and the feature vectors of the physiological index nodes are composed of heart rate, blood pressure, and blood oxygen data collected by the multi-functional physiological monitoring bracelet; motion state nodes, and the feature vectors of the motion state nodes are composed of motion data collected by the body movement sensor and the pedometer; blood glucose level nodes, and the feature vectors of the blood glucose level nodes are composed of blood glucose data collected by the intelligent blood glucose meter;

[0029] The edge relationship layer includes: temporal association edges connecting different temporal data of the same type of nodes; multi-modal association edges connecting different types of nodes at the same moment; causal association edges based on the causal relationships between different nodes established by the expert knowledge base;

[0030] The attention mechanism layer includes: a temporal attention module for assigning weights to data in different time windows; a node attention module for assigning importance weights to different types of nodes; and an edge attention module for assigning weights to different types of associated edges.

[0031] Preferably, the health risk assessment model further includes:

[0032] A graph convolution module for extracting local association features between nodes;

[0033] A temporal convolution module for extracting temporal pattern features;

[0034] A risk prediction module for generating multi-dimensional health risk scores based on the extracted features.

[0035] Preferably, the exercise plan optimization model constructs an exercise feature library for different disease types based on combining an expert knowledge base, including exercise types, intensity ranges, durations, and taboos, providing a basic decision-making basis for the exercise model;

[0036] Extract the patient's physiological indicators and exercise parameters from real-time data as input features for exercise assessment; use a deep neural network model to match and analyze the extracted features with exercise constraint rules, calculate the safety factor and effectiveness score of the current exercise state, and achieve real-time exercise risk assessment.

[0037] Preferably, based on the life rhythm management strategy, the expert knowledge base constructs a sleep feature library for different disease types, including recommended sleep durations, bedtime windows, sleep stage ratios, and sleep quality standards, as an evaluation benchmark; obtain the patient's sleep data, and use a fuzzy inference system to calculate the compliance degrees of each dimension of sleep, including duration compliance rate, regularity score, and quality score, and finally generate a sleep score from 0 to 100 by weighting.

[0038] Preferably, giving the work and rest advice plan specifically includes:

[0039] According to the sleep score, combined with the patient's living habits and treatment plan, generate personalized work and rest advice, including recommended work and rest times, sleep environment adjustment, and behavior intervention measures.

[0040] Another aspect of the present invention provides a home care management method for chronic disease patients, including the following steps:

[0041] S1. Obtain the patient's real-time data through the terminal perception module: collect physiological indicators and exercise data using a monitoring bracelet; collect blood glucose data using a blood glucose meter; collect sleep data using a sleep sensor;

[0042] S2. The edge computing module uses the home gateway device to collect real-time data from the terminal sensing module, preprocesses the real-time data through the edge intelligent analysis system, and realizes real-time monitoring and early warning with fast response locally;

[0043] S3. The data processing center of the cloud service module processes data: uses a distributed storage system to store multi-source data; uses a data cleaning and standardization processing module to process the original data; uses a multi-source data fusion analysis module based on deep learning to fuse and analyze data;

[0044] S4. The intelligent decision-making system of the cloud service module generates personalized solutions: uses a health risk assessment model to evaluate the health status of patients; uses an exercise plan optimization model to generate exercise plans; uses a work and rest advice generation model to formulate work and rest advice plans;

[0045] S5. The optimization module comprehensively coordinates the generated comprehensive risk index, exercise plan, and work and rest advice plan; dynamically adjusts the plan parameters according to patient feedback and implementation conditions; pushes the optimized personalized plan to the user terminal for execution; continuously tracks the execution effect of the plan and performs iterative optimization.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] The present invention proposes a home care management platform for chronic disease patients. Through the health risk assessment model, the multi-layer temporal heterogeneous graph neural network can capture the complex associations between different types of data. The weight assignment based on the attention mechanism improves the accuracy of the assessment, and the multi-dimensional risk score provides a comprehensive health status assessment; the exercise plan optimization model ensures exercise safety based on the constraint conditions of the patient model; real-time monitoring and evaluation provide timely exercise guidance and early warning; the reinforcement learning method realizes the continuous optimization of the plan; the work and rest advice generation model provides a personalized work and rest advice plan, improving the executability of the plan. The work and rest time arrangement considering the disease characteristics is more scientific, and the dynamic evaluation and adjustment improve the effect of work and rest management. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a structural diagram of a home care management platform for chronic disease patients provided by the present invention.

[0049] Figure 2 It is a flowchart of a home care management method for chronic disease patients provided by the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0050] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0051] Figure 1 The figure is a structural diagram of a home care management platform for chronic disease patients provided by the present invention. An embodiment of the present invention provides a home care management platform for chronic disease patients, as Figure 1 shown, including:

[0052] A terminal perception module, including a monitoring bracelet and a blood glucose meter, for obtaining real-time data of patients; the monitoring bracelet is used to monitor the physiological indicators and exercise data of patients in real time, and the monitoring bracelet integrates a heart rate sensor, a blood pressure sensor, a blood oxygen sensor, a body movement sensor and a pedometer for collection; the blood glucose meter is used to monitor the blood glucose data of patients in real time, and the blood glucose meter supports NFC wireless transmission function;

[0053] An edge computing module, including a home gateway device and an edge intelligent analysis system; the home gateway device is used to collect the real-time data of the terminal perception module, and the home gateway device includes a communication module integrating Bluetooth, WiFi and ZigBee communication protocols;

[0054] A cloud service module, the cloud service module includes a data processing center and an intelligent decision-making system; the data processing center includes a distributed storage system, a data cleaning and standardization processing module, and a multi-source data fusion analysis module based on deep learning; the intelligent decision-making system includes a health risk assessment model, an exercise plan optimization model, and an expert knowledge base including exercise guidance strategies and life rhythm management strategies; the health risk assessment model uses a graph structure analysis model constructed based on a graph neural network to obtain a multi-dimensional health risk score, and calculates a comprehensive risk index based on the scores of each dimension; the exercise plan optimization model constructs and applies an exercise model according to physiological indicators and real-time data, evaluates the current exercise score, and gives an exercise plan in combination with the exercise guidance strategy of the expert knowledge base;

[0055] An optimization module, which comprehensively coordinates the generated results, dynamically adjusts the parameters of the exercise plan according to the patient's feedback and execution situation, and pushes the optimized personalized plan to the user terminal for execution.

[0056] In an embodiment of the present invention, further, the monitoring bracelet further includes a sleep sensor, and the sleep sensor is used to monitor the sleep data of the patient in real time;

[0057] The intelligent decision-making system further includes a work and rest advice generation model. The work and rest advice generation model evaluates the current sleep score based on the disease type and sleep data, and gives a work and rest advice plan in combination with the life rhythm management strategy in the expert knowledge base.

[0058] In an implementation manner of the present invention, the edge intelligent analysis system is used for real-time data preprocessing to achieve fast-response real-time monitoring and early warning locally, including:

[0059] The data access layer is used for receiving real-time data, including:

[0060] The multi-channel data cache pool is used for receiving and real-time data in parallel;

[0061] The data frame synchronization unit is used for time stamp alignment of real-time data from different sensors;

[0062] The data quality evaluation unit is used for filtering abnormal data frames;

[0063] The real-time calculation layer includes:

[0064] The feature extraction and processing unit includes: a parallel Fourier transform module for extracting the frequency domain features of physiological signals; a sliding window statistics module for calculating the time domain statistical features; a multi-scale wavelet decomposition module for extracting the multi-scale features of signals;

[0065] The fast detection and processing unit includes: a signal prediction module based on Kalman filtering; an anomaly detection module based on a lightweight neural network; a threshold determination module based on a rule engine;

[0066] The early warning decision layer adopts a hierarchical processing mechanism, including:

[0067] The lightweight early warning evaluation unit includes: a fast classification module based on a decision tree; a risk scoring module based on fuzzy logic; an early warning level determination module based on a rule base;

[0068] The early warning trigger control unit includes: an early warning priority sorting module; an early warning information merging module; an early warning sending control module.

[0069] Among them, the edge intelligent analysis system achieves fast response in the following ways: adopting a pipeline parallel processing architecture, where each layer processes data at different times simultaneously; using lightweight algorithm models to ensure that the single processing delay does not exceed 100 milliseconds; implementing hierarchical caching of data and adopting different processing strategies for data with different priorities; maintaining a dynamically updated personalized parameter library locally for fast anomaly determination.

[0070] In an embodiment of the present invention, the distributed storage system adopts a multi-node hierarchical storage architecture, stores real-time data in cache nodes, and distributes historical data to storage nodes with different performance levels according to time spans, realizing efficient access to data;

[0071] The data cleaning and standardization processing module adopts a pipeline processing method, and sequentially performs outlier detection, missing value filling, time alignment, and numerical normalization to ensure data quality and consistency;

[0072] The multi-source data fusion and analysis module adopts a deep learning model based on the attention mechanism to extract and fuse features of real-time data from different sources to obtain corresponding feature vectors.

[0073] In an embodiment of the present invention, the health risk assessment model is a multi-layer temporal heterogeneous graph neural network model, including:

[0074] A node feature layer, including: a physiological index node, where the physiological index node feature vector is composed of heart rate, blood pressure, and blood oxygen data collected by the multi-functional physiological monitoring bracelet; a motion state node, where the motion state node feature vector is composed of motion data collected by the body movement sensor and the pedometer; a blood glucose level node, where the blood glucose level node feature vector is composed of blood glucose data collected by the intelligent blood glucose meter;

[0075] An edge relationship layer, including: a temporal association edge, connecting different temporal data of the same type of node; a multi-modal association edge, connecting different types of nodes at the same moment; a causal association edge, a causal relationship between different nodes established based on the expert knowledge base;

[0076] An attention mechanism layer, including: a temporal attention module, used to assign weights to data in different time windows; a node attention module, used to assign importance weights to different types of nodes; an edge attention module, used to assign weights to different types of association edges.

[0077] Further, in an embodiment of the present invention, the health risk assessment model further includes:

[0078] A graph convolution module, used to extract local association features between nodes;

[0079] A temporal convolution module, used to extract temporal pattern features;

[0080] A risk prediction module, generating a multi-dimensional health risk score based on the extracted features.

[0081] Specific implementation steps of the graph convolution module: First, construct a multi-layer graph convolution network, where each layer contains multiple convolutional kernels for capturing node association patterns at different scales. For each node, generate a local structure representation of the node by aggregating the feature information of its neighbor nodes. During the feature aggregation process, use the attention mechanism to weight the contributions of different neighbor nodes to highlight the influence of important nodes. By stacking multiple layers of graph convolution, gradually expand the receptive field to achieve the extraction of deep association features. Finally, merge the feature representations of all nodes to obtain an association feature vector reflecting the overall network structure.

[0082] Specific implementation steps of the temporal convolution module: Adopt a one-dimensional convolutional neural network structure and design multiple convolutional layers with different convolutional kernel sizes for capturing change patterns at different time scales. When performing convolutional operations, use causal convolution to ensure that the prediction at the current moment only depends on historical data. Expand the receptive field through dilated convolution technology to achieve the modeling of long-term temporal dependencies. Combine residual connections and layer normalization to improve the convergence and generalization ability of the model. Finally, output a feature vector reflecting temporal change features.

[0083] Specific implementation steps of the risk prediction module: Integrate the local association features extracted by the graph convolution module and the temporal pattern features extracted by the temporal convolution module to construct a unified feature representation. Use a multi-layer perceptron to perform non-linear transformation on the integrated features and map them to the scoring space of multiple risk dimensions. During the scoring generation process, combine the risk assessment criteria in the expert knowledge base to ensure the interpretability and reliability of the scoring results. For each risk dimension, output a standardized score from 0 to 100, and generate corresponding risk levels and warning messages. Finally, calculate the comprehensive risk index based on the scores of each dimension to provide a reference basis for clinical decision-making.

[0084] In one implementation of the present invention, the exercise plan optimization model constructs an exercise feature library for different disease types based on the combined expert knowledge base, including exercise types, intensity ranges, durations, and taboos, providing a basic decision-making basis for the exercise model;

[0085] Extract the patient's physiological indicators and exercise parameters from real-time data as input features for exercise assessment; use a deep neural network model to match and analyze the extracted features with exercise constraint rules, calculate the safety factor and effect score of the current exercise state, and achieve real-time exercise risk assessment.

[0086] In an embodiment of the present invention, based on the life rhythm management strategy, the expert knowledge base constructs a sleep feature library for different disease types, including the recommended sleep duration, bedtime window, sleep stage ratio, and sleep quality standard, as the evaluation benchmark; obtains the patient's sleep data, and uses the fuzzy inference system to calculate the compliance degree of each dimension of sleep, including the duration compliance rate, regularity score, and quality score, and finally generates a sleep score from 0 to 100 through weighted calculation.

[0087] Further, in an embodiment of the present invention, the specific content of the work and rest advice plan includes:

[0088] According to the sleep score, combined with the patient's living habits and treatment plan, personalized work and rest advice is generated, including the recommended work and rest time, sleep environment adjustment, and behavior intervention measures.

[0089] Figure 2 This is a flowchart of a home care management method for chronic disease patients provided by the present invention. As Figure 2 shown, an embodiment of the present invention also provides a home care management method for chronic disease patients, including the following steps:

[0090] S1. Obtain the patient's real-time data through the terminal perception module: use the monitoring bracelet to collect physiological indicators and exercise data; use the blood glucose meter to collect blood glucose data; use the sleep sensor to collect sleep data;

[0091] S2. The edge computing module uses the home gateway device to collect the real-time data of the terminal perception module, and preprocesses the real-time data through the edge intelligent analysis system to achieve real-time monitoring and early warning with fast response locally;

[0092] S3. Perform data processing through the data processing center of the cloud service module: use the distributed storage system to store multi-source data; use the data cleaning and standardization processing module to process the original data; use the multi-source data fusion analysis module of deep learning to fuse and analyze the data;

[0093] S4. Generate a personalized plan through the intelligent decision-making system of the cloud service module: use the health risk assessment model to evaluate the patient's health status; use the exercise plan optimization model to generate an exercise plan; use the work and rest advice generation model to formulate a work and rest advice plan;

[0094] S5. The optimization module comprehensively coordinates the generated comprehensive risk index, exercise plan, and work and rest advice plan; dynamically adjusts the plan parameters according to the patient's feedback and implementation situation; pushes the optimized personalized plan to the user terminal for execution; continuously tracks the execution effect of the plan and performs iterative optimization.

[0095] In step S1 of the present invention, the physiological indicators and exercise data of patients are comprehensively monitored through a monitoring bracelet, a blood glucose meter, and a sleep sensor, ensuring the continuity and integrity of real-time data collection. The real-time data collection from different dimensions enables medical staff to comprehensively understand the health status of patients, facilitating the early detection of potential problems. The use of intelligent devices for automatic collection reduces the burden on patients for manual recording and improves data accuracy.

[0096] In step S2 of the present invention, local data preprocessing is performed through an edge computing module, reducing the data transmission burden and improving the system response speed. Quick warnings can be realized locally to detect and respond to abnormal situations of patients in a timely manner. The home gateway uniformly manages multiple sensing devices, simplifying the data collection process and improving system stability.

[0097] In step S3 of the present invention, distributed storage ensures the secure storage and fast access of a large amount of patient data. Data cleaning and standardization processing improve data quality, providing a reliable basis for subsequent analysis. The multi-source data fusion analysis based on deep learning can discover health trends and correlations that are difficult to find with a single data source.

[0098] The health risk assessment model in step S4 of the present invention can predict potential health problems and achieve preventive intervention. The exercise plan optimization model formulates a scientific exercise plan according to the actual situation of patients to ensure safety and effectiveness. The daily routine advice generation model helps patients establish healthy living habits and improve the management effect of chronic diseases.

[0099] The comprehensive coordination in step S5 of the present invention ensures the mutual cooperation between various suggestions and avoids conflicts. The dynamic adjustment according to patient feedback improves the executability and pertinence of the plan. Continuous tracking and iterative optimization ensure that the treatment plan can be adjusted in a timely manner as the patient's condition changes to maintain the best effect.

[0100] Working principle: Various physiological indicators and exercise data of patients are collected through terminal devices such as a monitoring bracelet, a blood glucose meter, and a sleep sensor, and then the data is aggregated and preprocessed by edge computing through a home gateway device to achieve local quick warnings. The preprocessed data is transmitted to the cloud, stored using a distributed storage system, and the data quality is ensured through data cleaning and standardization processing. Then, multi-source data fusion analysis is performed using deep learning algorithms. Based on the analysis results, the intelligent decision-making system uses a health risk assessment model, an exercise plan optimization model, and a daily routine advice generation model to formulate a personalized health management plan. Finally, through continuous tracking of plan execution and patient feedback, the plan is dynamically optimized and adjusted, thereby realizing the precise management of chronic disease patients.

[0101] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art will appreciate that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A home care management platform for chronic disease patients, characterized in that including: The terminal perception module includes a monitoring bracelet and a blood glucose meter, and is used to obtain real-time data of patients. The monitoring bracelet is used to monitor the physiological indicators and exercise data of patients in real time. The monitoring bracelet integrates a heart rate sensor, a blood pressure sensor, a blood oxygen sensor, a body movement sensor and a pedometer for collection. The blood glucose meter is used to monitor the blood glucose data of patients in real time, and the blood glucose meter supports NFC wireless transmission function. The edge computing module includes a home gateway device and an edge intelligent analysis system. The home gateway device is used to collect the real-time data of the terminal perception module. The home gateway device includes a communication module integrating Bluetooth, WiFi and ZigBee communication protocols. The cloud service module includes a data processing center and an intelligent decision-making system. The data processing center includes a distributed storage system, a data cleaning and standardization processing module, and a multi-source data fusion analysis module based on deep learning. The intelligent decision-making system includes a health risk assessment model, an exercise plan optimization model, and an expert knowledge base including exercise guidance strategies and life rhythm management strategies. The health risk assessment model uses a graph structure analysis model constructed based on a graph neural network to obtain a multi-dimensional health risk score, and calculates a comprehensive risk index based on the scores of each dimension. The exercise plan optimization model constructs and applies an exercise model according to physiological indicators and real-time data, evaluates the current exercise score, and gives an exercise plan in combination with the exercise guidance strategy of the expert knowledge base. The optimization module comprehensively coordinates the generated results, dynamically adjusts the exercise plan parameters according to the patient's feedback and execution situation, and pushes the optimized personalized plan to the user terminal for execution.

2. The home care management platform for chronic disease patients according to claim 1, wherein The monitoring bracelet also includes a sleep sensor, and the sleep sensor is used to monitor the sleep data of patients in real time. The intelligent decision-making system also includes a work and rest suggestion generation model. The work and rest suggestion generation model evaluates the current sleep score according to the disease type and sleep data, and gives a work and rest suggestion plan in combination with the life rhythm management strategy of the expert knowledge base.

3. The home care management platform for chronic disease patients according to claim 1, characterized in that, The edge intelligent analysis system is used for real-time data preprocessing, and realizes fast-response real-time monitoring and early warning locally, including: The data access layer is used for real-time data reception, including: The multi-channel data cache pool is used for parallel reception and real-time data. The data frame synchronization unit is used to align the timestamps of the real-time data of different sensors. The data quality evaluation unit is used to filter abnormal data frames. The real-time calculation layer includes: The feature extraction processing unit includes: a parallel Fourier transform module for extracting the frequency domain features of physiological signals; a sliding window statistics module for calculating the time domain statistical features; a multi-scale wavelet decomposition module for extracting the multi-scale features of signals. The fast detection processing unit includes: a signal prediction module based on Kalman filtering; an anomaly detection module based on a lightweight neural network; a threshold determination module based on a rule engine. The early warning decision layer adopts a hierarchical processing mechanism, including: The lightweight early warning evaluation unit includes: a fast classification module based on a decision tree; a risk scoring module based on fuzzy logic; an early warning level determination module based on a rule base. The early warning trigger control unit includes: an early warning priority sorting module; an early warning information merging module; an early warning sending control module.

4. The home care management platform for chronic disease patients according to claim 1, characterized in that The distributed storage system adopts a multi-node hierarchical storage architecture, stores real-time data in cache nodes, and distributes historical data to storage nodes with different performance levels according to the time span, realizing efficient access to data; The data cleaning and standardization processing module adopts a pipeline processing method, and sequentially performs outlier detection, missing value filling, time alignment, and numerical normalization to ensure data quality and consistency; The multi-source data fusion analysis module adopts a deep learning model based on the attention mechanism to extract and fuse features of real-time data from different sources to obtain corresponding feature vectors.

5. The home care management platform for chronic disease patients according to claim 1, wherein The health risk assessment model is a multi-layer temporal heterogeneous graph neural network model, including: The node feature layer includes: a physiological index node, and the physiological index node feature vector is composed of heart rate, blood pressure, and blood oxygen data collected by the multi-functional physiological monitoring bracelet; a motion state node, and the motion state node feature vector is composed of motion data collected by the body motion sensor and the pedometer; a blood glucose level node, and the blood glucose level node feature vector is composed of blood glucose data collected by the intelligent blood glucose meter. The edge relationship layer includes: a temporal association edge, connecting different temporal data of the same type of node; a multi-modal association edge, connecting different types of nodes at the same moment; a causal association edge, establishing a causal relationship between different nodes based on the expert knowledge base. The attention mechanism layer includes: a temporal attention module, used to assign weights to data in different time windows; a node attention module, used to assign importance weights to different types of nodes; an edge attention module, used to assign weights to different types of association edges.

6. The home care management platform for chronic disease patients according to claim 1, wherein The health risk assessment model also includes: A graph convolution module, used to extract local association features between nodes; A temporal convolution module, used to extract temporal pattern features; A risk prediction module, generating multi-dimensional health risk scores based on the extracted features.

7. The home care management platform for chronic disease patients according to claim 1, characterized in that, The exercise plan optimization model constructs an exercise feature library for different disease types based on the combination of the expert knowledge base, including exercise type, intensity range, duration, and taboos, providing a basic decision-making basis for the exercise model; Extract the patient's physiological indicators and exercise parameters from the real-time data as input features for exercise evaluation; adopt a deep neural network model to match and analyze the extracted features with exercise constraint rules, calculate the safety factor and effect score of the current exercise state, and realize real-time exercise risk assessment.

8. The home care management platform for chronic disease patients according to claim 2, characterized in that Based on the life rhythm management strategy, the expert knowledge base constructs a sleep feature library for different disease types, including recommended sleep duration, bedtime window, sleep stage ratio, and sleep quality standard, as the evaluation benchmark; obtain the patient's sleep data, and use the fuzzy inference system to calculate the compliance degree of each dimension of sleep, including duration compliance rate, regularity score, and quality score, and finally generate a sleep score of 0-100 after weighting.

9. The home care management platform for chronic disease patients according to claim 8, characterized in that, The specific content of the work and rest advice plan includes: According to the sleep score, combined with the patient's living habits and treatment plan, generate personalized work and rest advice, including recommended work and rest time, sleep environment adjustment, and behavior intervention measures.

10. A home care management method for chronic disease patients, characterized in that, Including the following steps: S1. Obtain the real-time data of the patient through the terminal perception module: collect physiological indicators and exercise data using a monitoring bracelet; collect blood glucose data using a blood glucose meter; collect sleep data using a sleep sensor; S2. The edge computing module uses a home gateway device to collect the real-time data of the terminal perception module, preprocess the real-time data through the edge intelligent analysis system, and achieve real-time monitoring and early warning with fast response locally; S3. Perform data processing through the data processing center of the cloud service module: store multi-source data using a distributed storage system; process the original data using a data cleaning and standardization processing module; fuse and analyze the data using a multi-source data fusion analysis module based on deep learning; S4. Generate a personalized plan through the intelligent decision-making system of the cloud service module: evaluate the patient's health status using a health risk assessment model; generate an exercise plan using an exercise plan optimization model; Generate a work and rest advice plan using a work and rest advice generation model; S5. The optimization module comprehensively coordinates the generated comprehensive risk index, exercise plan, and work and rest advice plan; Dynamically adjust the plan parameters according to the patient's feedback and implementation situation; Push the optimized personalized plan to the user terminal for execution; Continuously track the implementation effect of the plan and perform iterative optimization.

Images