A Community-Based Precision Care Follow-up Method and System for Chronic Diseases

Abstract

This invention relates to the field of healthcare management technology, and more particularly to a method and system for precise follow-up care of chronic diseases in communities. The method includes the following steps: acquiring user vital sign data and user lifestyle data; fitting the user's vital sign data with a pre-defined chronic disease medical atlas to obtain user chronic disease index data; adjusting the user's chronic disease index based on the user's lifestyle data and chronic disease index data to obtain adjusted chronic disease index data; acquiring user location data corresponding to the user's vital sign data, and dividing the community based on the user's location data and adjusted chronic disease index data to obtain user chronic disease index community division data; and constructing a follow-up path based on the user chronic disease index community division data to obtain community chronic disease follow-up plan data for conducting community chronic disease nursing follow-up work. This invention improves the efficiency and accuracy of community chronic disease nursing follow-up and reduces medical costs.

Description

Technical Field

[0001] This invention relates to the field of healthcare management technology, and in particular to a method and system for precise nursing follow-up of chronic diseases in the community. Background Technology

[0002] Chronic diseases (such as diabetes, hypertension, and cardiovascular diseases) have become a major public health problem worldwide. With the advent of an aging society and changes in lifestyle, the number of chronic disease patients in communities is constantly rising, leading to a sharp increase in demand for medical resources and posing significant challenges to primary healthcare institutions and public health systems. However, in traditional chronic disease management models, uneven distribution of medical resources and inflexible nursing follow-up plans result in unsatisfactory nursing outcomes and ineffective monitoring and management of patients' long-term health status. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a method and system for precise nursing follow-up of chronic diseases in the community, thereby resolving at least one of the aforementioned technical issues.

[0004] This application provides a method for precise follow-up care of chronic diseases in the community, including the following steps:

[0005] Step S1: Obtain user vital signs data and user lifestyle data;

[0006] Step S2: Fit the user's vital signs data and the preset chronic disease medical atlas to obtain the user's chronic disease index data; adjust the user's chronic disease index based on the user's lifestyle data and the user's chronic disease index data to obtain the user's chronic disease index adjustment data.

[0007] Step S3: Obtain the user location data corresponding to the user's vital signs data, and divide the community according to the user location data and the user's chronic disease index adjustment data to obtain the user's chronic disease index community division data.

[0008] Step S4: Construct follow-up pathways based on the community segmentation data of users' chronic disease index to obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up work.

[0009] This invention, by acquiring users' vital sign and lifestyle data, allows the system to adjust nursing plans based on each user's specific chronic disease index. This method enables real-time adjustments based on users' chronic disease index and lifestyle data. Changes in lifestyle and vital signs directly affect the calculation of the chronic disease index, and adjustments to the chronic disease index update follow-up pathways and plans. By combining user location data with chronic disease index adjustment data, this method delineates community units with high and low incidence of chronic diseases, enabling better allocation of community resources, concentrating efforts on caring for high-risk patients, optimizing the use of community medical resources, and improving the efficiency of medical resources. Through precise community delineation and rational arrangement of follow-up pathways, this method can effectively reduce medical costs. High-risk patients will receive priority medical care, while the follow-up frequency for low-risk patients can be appropriately reduced, decreasing unnecessary resource consumption and reducing the overall medical cost burden.

[0010] Preferably, step S1 specifically includes:

[0011] Step S11: Collect user vital signs data and user movement data through a portable device, wherein the user vital signs data includes user heart rate data and user blood pressure data;

[0012] Step S12: Extract user lifestyle features based on user heart rate data and user exercise data to obtain user lifestyle feature data;

[0013] Step S13: Obtain historical user lifestyle data and historical user vital sign data;

[0014] Step S14: Perform lifestyle clustering calculations on historical user vital sign data based on historical user lifestyle data to obtain historical user lifestyle clustering feature data;

[0015] Step S15: Perform fitting mapping based on user lifestyle feature data and historical user lifestyle clustering feature data to obtain user lifestyle feature fitting mapping data;

[0016] Step S16: Obtain the user's exercise time data corresponding to the user's exercise data;

[0017] Step S17: Evaluate and filter the user lifestyle characteristics fitting mapping data based on the user exercise time data to obtain user lifestyle data.

[0018] This invention extracts lifestyle characteristics based on user heart rate and exercise data, revealing behavioral features such as activity patterns, exercise frequency, and intensity. Lifestyle clustering features are obtained through cluster analysis of historical user lifestyle and vital sign data. Users continuously and automatically collect vital sign data, including key physiological indicators such as heart rate and blood pressure, and exercise data (such as exercise type, intensity, and steps), using portable devices (e.g., smart bracelets, smartwatches). By analyzing heart rate and exercise data, the system can identify user exercise intensity, type, and habits, accurately reflecting their true lifestyle. The system retrieves lifestyle and vital sign data from historical databases of related or similar user groups as a benchmark to establish a comparative model. Clustering calculations are performed on historical user lifestyle data to form multiple lifestyle types (e.g., active, sedentary, balanced), and features of each lifestyle type are extracted. The user's lifestyle feature data is fitted with historical clustering results to generate user lifestyle feature fitting mapping data. Through matching analysis, the user's lifestyle type is identified. The system uses users' exercise time data to fit and map their lifestyle characteristics, then further filters and evaluates the data to generate accurate lifestyle data.

[0019] Preferably, the lifestyle clustering calculation specifically involves:

[0020] Based on historical user lifestyle data, spectral clustering calculations are performed on historical user vital sign data to obtain the first historical user lifestyle cluster feature data.

[0021] Based on historical user lifestyle data, graph clustering calculations are performed on historical user vital sign data to obtain second historical user lifestyle clustering feature data.

[0022] The first historical user lifestyle cluster feature data and the second historical user lifestyle cluster feature data are mapped and aligned to obtain historical user lifestyle feature aligned data.

[0023] Similarity calculations and weighted merging are performed on historical user lifestyle feature alignment data to obtain historical user lifestyle cluster feature data.

[0024] This invention utilizes different algorithms to cluster historical user data. Spectral clustering is suitable for handling complex relationships in high-dimensional data, while graph clustering is better suited for handling network data with complex structures. Aligning the feature data generated by different clustering algorithms ensures that features from different methods can be compared and integrated under the same criteria. Similarity calculations using the aligned data identify common characteristics and key behavioral patterns among historical users. Weighted merging further integrates the advantages of different algorithms, improving the representativeness and accuracy of the feature data. Through clustering calculations and data merging, this method enables in-depth analysis of the lifestyle characteristics of historical users, providing a more detailed assessment. Employing different clustering methods (spectral clustering and graph clustering) can mitigate the bias that may arise from a single algorithm. By weighted merging of the clustering results, the advantages of different methods can be effectively integrated, reducing potential biases caused by different algorithm choices and improving the overall accuracy of data analysis.

[0025] Preferably, the spectral clustering calculation specifically involves:

[0026] A graph is constructed based on historical user lifestyle data and historical user physical characteristics data to obtain historical user lifestyle and physical characteristics association graph data.

[0027] Based on the historical user lifestyle characteristics correlation map data, similarity map feature encoding is performed to obtain historical user lifestyle characteristics map feature data;

[0028] The Laplace matrix is ​​calculated based on the historical user lifestyle characteristic data to obtain the first historical user lifestyle characteristic matrix data.

[0029] Based on the first historical user lifestyle characteristic matrix data, feature value decomposition and embedding space generation are performed to obtain historical lifestyle characteristic feature embedding space data.

[0030] Clustering is performed on spatial data embedded with historical lifestyle characteristics to obtain the first historical user lifestyle clustering characteristic data.

[0031] This invention constructs a correlation graph from historical users' lifestyle and vital sign data, comprehensively capturing multi-dimensional relationships between users. The encoding process preserves the graph's topology and relationships between nodes, ensuring no important information is lost during conversion. The Laplace matrix reflects the graph's global structural features, including node connectivity and overall topology. Eigenvalue decomposition maps high-dimensional lifestyle and vital sign data to a low-dimensional embedding space, reducing data dimensionality and computational complexity. In the embedding space, key features are preserved, while noise and redundant information are filtered out, improving data quality. Clustering within the embedding space more accurately groups users with similar lifestyle and vital sign characteristics. The resulting first historical user lifestyle clustering feature data helps identify commonalities and differences among different user groups, providing a basis for follow-up health management.

[0032] Preferably, the similarity graph feature encoding is specifically as follows:

[0033] Multi-scale convolution calculations are performed based on historical user lifestyle characteristic correlation graph data to obtain correlation graph feature convolution layer data;

[0034] Global and local multi-head self-attention calculation is performed on the convolutional layer data of the association graph features to obtain the first association graph feature self-attention data;

[0035] Perform masked multi-head self-attention calculation on the convolutional layer data of the association graph to obtain the feature self-attention data of the second association graph.

[0036] Pooling layer calculation is performed on the self-attention data of the first association graph feature to obtain the association graph feature pooling layer data, and feature distillation is performed on the self-attention data of the second association graph feature to obtain the association graph feature distillation data.

[0037] Based on the feature pooling layer data and feature distillation data of the association graph, a similar graph is constructed to obtain the feature data of historical user lifestyle characteristics.

[0038] This invention utilizes feature data with strong representational capabilities to construct more accurate similarity graphs, providing a solid foundation for subsequent clustering analysis and improving the accuracy of chronic disease risk assessment and care plan development. Through multiple attention mechanisms and feature distillation, the model can better cope with noise and missing data, improving adaptability to new data and predictive performance. Precise feature encoding and similarity graph construction enable a deeper understanding of users' lifestyles and health status, thus supporting customized care interventions and follow-up plans. Convolutional operations for graph-structured data are suitable for processing data in non-Euclidean spaces, improving the ability to model complex network relationships. Global-local multi-head self-attention can simultaneously consider the local neighborhood information and global structural information of nodes in the graph, deeply understanding the complex relationships in user data. By masking part of the input during training, the model can learn more robust feature representations, reducing overfitting. Pooling operations can simplify feature representations, retain the most important information, reduce computational complexity, and prevent overfitting. Feature distillation extracts knowledge from complex models and transfers it to simplified models, achieving model lightweighting, improving computational efficiency, and maintaining performance. By combining the features of pooling and distillation, the model can leverage feature information from different levels and sources to enhance the understanding of users' health status. Similarity graphs help to more accurately identify user groups, improve the performance of clustering algorithms, and thus enhance the effectiveness of community chronic disease care follow-up programs.

[0039] Preferably, the graph clustering calculation specifically involves:

[0040] A similarity matrix is ​​constructed based on historical user lifestyle data and historical user physical characteristics data to obtain historical user lifestyle characteristics similarity matrix data.

[0041] Based on the historical user lifestyle characteristics similarity matrix data, an adjacency matrix and a degree matrix are constructed to obtain historical user lifestyle characteristics adjacency matrix data and historical user lifestyle characteristics degree matrix data, respectively.

[0042] The Laplace matrix is ​​calculated based on the historical user lifestyle characteristic adjacency matrix data and the historical user lifestyle characteristic degree matrix data to obtain the first historical user lifestyle characteristic matrix data.

[0043] Based on the first historical user lifestyle characteristic matrix data, feature value decomposition and graph embedding are performed to generate historical user lifestyle characteristic embedding data.

[0044] Hierarchical clustering is performed based on the embedded data of historical user lifestyle characteristics to obtain the second historical user lifestyle clustering feature data.

[0045] The graph clustering method in this invention can handle nonlinear, high-dimensional, and heterogeneous data, exhibiting strong adaptability in complex user data scenarios and making it suitable for community needs analysis in precision care for chronic diseases. By calculating a similarity matrix, the similarity between users is accurately measured, helping to identify user groups with similar lifestyles and physical characteristics, thus facilitating the development of more precise care plans that conform to the general characteristics of community users. The similarity matrix can uncover implicit associations from different dimensions of lifestyle and physical characteristic data, helping to discover potential health association models between users or between users and lifestyles. The adjacency matrix constructs relationships between users through threshold filtering, and the generated graph structure reflects the global connectivity of users. The combination of the degree matrix and the adjacency matrix provides necessary topological information for subsequent Laplace matrix calculation. Eigenvalue decomposition effectively reduces dimensionality, simplifying high-dimensional complex data into a low-dimensional embedding space while retaining important feature information and avoiding information loss. The data in the embedding space can more clearly reflect the feature similarity of user groups, facilitating more accurate clustering analysis and improving clustering results. Hierarchical clustering can capture the hierarchical structure within a user group and identify user groups at different granularities. It is particularly suitable for identifying subgroups of users with different potential needs for chronic diseases.

[0046] Preferably, the evaluation and screening specifically includes:

[0047] Based on user exercise time data, predict user biological clock behavior to obtain user biological clock behavior data;

[0048] Based on user exercise data and user biological clock behavior data, indoor and outdoor lifestyles are filtered to obtain user indoor and outdoor lifestyle filtering data.

[0049] User lifestyle data is obtained by fitting data based on user internal and external lifestyle filtering data and user lifestyle feature fitting mapping data.

[0050] In this invention, a portable device collects users' exercise time data. By analyzing the user's daily exercise start and end times, frequency, and duration, the system can accurately infer the user's daily routine based on this data. Combining user exercise data (such as GPS location, step count, indoor / outdoor detection data, etc.) with circadian rhythm data, the system can determine whether the user's exercise takes place outdoors or whether they spend most of their time indoors. The system identifies the ratio of indoor to outdoor activities: if most exercise is outdoors (e.g., walking, running), it is categorized as "primarily outdoor activity"; if most exercise occurs indoors (e.g., walking at home, working out in a gym), it is categorized as "primarily indoor activity." The system fits and maps the user's current indoor / outdoor lifestyle data with historical lifestyle data. By matching similar historical data, it predicts the user's future lifestyle trends and outputs lifestyle data. If the user's lifestyle highly matches the behavioral patterns of historically active users, the system classifies the user as active; if the user exhibits excessive indoor activity and a lack of outdoor activity, the system identifies the user as sedentary. The system maps users’ actual exercise data with historical users’ lifestyle characteristics data, matching similar patterns and features to obtain accurate user behavior data. Compared with interface input, it can better reflect the actual situation and provides an accurate data foundation for the development of follow-up plans.

[0051] Preferably, step S3 specifically includes:

[0052] Step S31: Obtain the user location data corresponding to the user's vital signs data;

[0053] Step S32: Correlate user location data and user chronic disease index adjustment data to obtain user chronic disease index location correlation data;

[0054] Step S33: Perform regional clustering calculations based on the location association data of the user's chronic disease index to obtain the location clustering feature data of the user's chronic disease index;

[0055] Step S34: Obtain community medical resource data;

[0056] Step S35: Match and divide the community medical resource data and the user chronic disease index location cluster feature data to obtain the user chronic disease index community division data.

[0057] This invention combines users' vital sign data with geographic location information to deeply analyze the distribution of users' chronic disease risks, helping medical institutions optimize the allocation of community nursing resources. Through regional clustering and medical resource matching, medical resources can be allocated efficiently and rationally within the community, ensuring that high-risk users receive priority care and improving the efficiency of community chronic disease management. By rationally matching medical resources with users' health conditions, this method can promote the balanced distribution of medical resources within the community, avoiding problems of resource concentration or uneven distribution. Through precise community segmentation, follow-up pathways can be optimized according to regional divisions, improving the efficiency of follow-up work and reducing follow-up costs. The community segmentation and resource matching steps enable medical institutions to flexibly adjust nursing strategies according to the health needs of users in each region, ensuring that all users receive nursing services suitable for their health conditions.

[0058] Preferably, step S4 specifically includes:

[0059] Step S41: Construct visitation routes based on user chronic disease index community segmentation data to obtain primary visitation route data;

[0060] Step S42: Prioritize the community division data based on the user's chronic disease index to obtain community division priority data;

[0061] Step S43: Obtain follow-up personnel data;

[0062] Step S44: Estimate follow-up resources based on follow-up personnel data to obtain follow-up resource data;

[0063] Step S45: Perform priority matching based on follow-up resource data and community division priority data to obtain community division follow-up matching data;

[0064] Step S46: Map the primary visit route data to the community-based follow-up matching data to obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up work.

[0065] This invention ensures the rational use of follow-up resources, improves follow-up efficiency, and reduces resource waste through preliminary path planning and prioritization. Based on users' chronic disease indices and community classifications, high-risk patients are prioritized to maximize follow-up effectiveness and avoid unnecessary visits. Combining user health status, community priority, and follow-up personnel capabilities, the most suitable follow-up plan is provided for each community, ensuring basic health management effectiveness while minimizing medical costs. The follow-up plan can be dynamically adjusted based on changes in resources, personnel, and patient health status, ensuring flexibility in follow-up work.

[0066] Preferably, this application also provides a community-based precision care follow-up system for chronic diseases, used to perform the community-based precision care follow-up method for chronic diseases as described above. The community-based precision care follow-up system for chronic diseases includes:

[0067] The user data collection module is used to acquire user vital sign data and user lifestyle data.

[0068] The user chronic disease index adjustment module is used to fit user vital sign data and a preset chronic disease medical atlas to obtain user chronic disease index data; and to adjust user chronic disease index based on user lifestyle data and user chronic disease index data to obtain user chronic disease index adjustment data.

[0069] The user chronic disease index community division module is used to obtain user location data corresponding to user vital sign data, and to divide the community based on user location data and user chronic disease index adjustment data to obtain user chronic disease index community division data.

[0070] The community chronic disease nursing follow-up pathway construction module is used to construct follow-up pathways based on the user's chronic disease index community division data, and obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up operations.

[0071] The beneficial effects of this invention are as follows: By fitting user vital sign data with a chronic disease medical atlas, the chronic disease risk of each patient can be more accurately assessed, forming a dynamic chronic disease index that reflects the trend of individual health status changes over time. The chronic disease index is adjusted in real time according to the user's lifestyle, ensuring that the care plan adapts to changes in the patient's actual health condition, thus making the care plan more flexible and dynamic, and effectively responding to sudden changes in the patient's lifestyle or health status. Through community segmentation, patient groups with similar health risks are identified and clustered, forming community health zones with different risk levels. This space-based chronic disease management method can effectively improve the allocation efficiency of community medical resources. By combining community priority segmentation, follow-up resource data, and follow-up personnel capabilities, the optimal follow-up path is dynamically generated, thereby significantly improving the efficiency and effectiveness of follow-up. This invention not only enables precise health management for individuals but also achieves broader community health management goals through cluster analysis and regional segmentation. Attached Figure Description

[0072] Other features, objects, and advantages of this application will become more apparent from the following detailed description of the non-limiting embodiments, taken with reference to the accompanying drawings:

[0073] Figure 1 A flowchart illustrating the steps of a community-based precision nursing follow-up method for chronic diseases is shown in one embodiment.

[0074] Figure 2 A flowchart illustrating the steps of a method for collecting data on chronic disease users in a community is shown in one embodiment.

[0075] Figure 3 A flowchart illustrating the steps of a user chronic disease index community segmentation method according to an embodiment is shown;

[0076] Figure 4 A flowchart illustrating the steps of a method for constructing a community chronic disease care follow-up pathway is shown in one embodiment. Detailed Implementation

[0077] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0078] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0079] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0080] A community health center in a certain city is managing a patient population suffering from chronic diseases such as diabetes and hypertension. The community has a population of approximately 5,000, of whom about 30% suffer from at least one chronic disease. With limited medical resources, how to efficiently manage and rationally allocate nursing resources has become a major issue.

[0081] Community healthcare institutions collect users' vital signs data, including heart rate, blood pressure, and blood glucose levels, through portable health devices (such as smart bracelets, blood pressure monitors, and blood glucose meters) and regular physical examinations. Users also record lifestyle data daily through smartphone applications, such as dietary habits, exercise time and type, and sleep duration. User A (45 years old, male) had an average blood pressure of 145 / 90 mmHg and a blood glucose level of 7.5 mmol / L over the past month, and exercised 3 times a week for 30 minutes each time. User B (60 years old, female) had a blood glucose level of 10.2 mmol / L, a high-sugar diet, and lack of exercise; the vital signs data indicated a high risk.

[0082] Based on users' vital signs (such as blood pressure and blood sugar) and lifestyle data (such as exercise and diet), the system fits the data to a pre-defined medical atlas for diabetes and hypertension, generating initial chronic disease index data. User A's chronic disease index is 60 (moderate risk), and User B's is 85 (high risk). Considering User A's regular exercise and healthy eating habits, the index is lowered; User B's index is further adjusted to 90 due to unhealthy diet and lack of exercise, indicating high risk. User A's index is adjusted from 60 to 55 because a healthy lifestyle reduces the risk of disease progression. User B's index rises to 90, indicating the need for urgent intervention.

[0083] Based on users' chronic disease index data and geographic location data, the system spatially divides and clusters users within the community, forming different chronic disease index communities. High-risk areas: These are areas where high-risk patients congregate. For example, the community where user B lives is identified as a high-risk area (40% of patients have a chronic disease index ≥80). Medium- and low-risk areas: For example, in the community where user A lives, 60% of patients have a chronic disease index below 60, and this area is classified as a medium- and low-risk area. Through regional clustering, it is possible to identify which communities require priority care.

[0084] Based on the chronic disease index categorized by community, the system generates a follow-up route. Simultaneously, it combines community priority data and the availability of follow-up personnel to perform resource matching and route optimization. The system prioritizes follow-up for patients in high-risk areas; for example, User B and other patients in their community are scheduled for follow-up this week. Based on the resources of community healthcare institutions and the capabilities of nursing staff, the system assigns experienced nurses to high-risk patient areas. The optimized follow-up route ensures that nurses can visit more patients with the shortest possible travel time, while focusing on high-risk communities. Route optimization also considers the working hours and traffic conditions of follow-up personnel.

[0085] The nursing team first visited the community where User B lived. User B was scheduled for a detailed interview and was advised to make adjustments to their diet and exercise. Other high-risk patients in the community were also given priority attention, and a detailed follow-up schedule was developed. User A lived in a medium-risk community, with a relatively low frequency of follow-up visits, including a routine check-up once a month. User A underwent routine blood pressure and blood sugar monitoring, and discussed their exercise habits with the nursing team, adjusting relevant recommendations accordingly. The system provides remote health monitoring for users in low-risk communities, allowing the nursing team to conduct home visits only when necessary, thus saving follow-up resources.

[0086] Based on the follow-up results, the system updates users' vital signs and lifestyle data in real time, and the chronic disease index is adjusted accordingly. User B's care plan was adjusted in real time. In the month following the follow-up, she followed dietary recommendations, her blood sugar levels decreased, and her chronic disease index dropped from 90 to 80. The system adjusted the follow-up frequency to once every two weeks. User A continued to maintain a good condition, and the index remained stable; the next follow-up was postponed to the following month.

[0087] The project's innovations lie in its ability to collect multi-dimensional user data (vital signs, lifestyle, and geographic location) and combine it with a medical atlas to generate and dynamically adjust chronic disease indices, thereby achieving personalized health management. By combining geographic location data and chronic disease indices, it enables precise community segmentation and cluster analysis, providing a scientific basis for regional health management. Intelligent planning algorithms, combined with community priorities and resource allocation, optimize follow-up pathways, improving follow-up efficiency and resource utilization while reducing unnecessary follow-up work. The follow-up plan is dynamically adjusted based on user health feedback, ensuring the flexibility and long-term effectiveness of nursing care and adapting to changes in patients' health conditions.

[0088] Please see Figures 1 to 4 This application provides a method for precision nursing follow-up of chronic diseases in the community, including the following steps:

[0089] Step S1: Obtain user vital signs data and user lifestyle data;

[0090] Specifically, users' physiological data, such as blood pressure, blood sugar, heart rate, and weight, are collected through portable devices (such as wearable devices and health checkup equipment). These devices can transmit the data to a medical data platform in real time via wireless networks. Lifestyle data is collected through questionnaires, applications, or health logs, including dietary habits, exercise frequency, sleep quality, and smoking and drinking habits.

[0091] Step S2: Fit the user's vital signs data and the preset chronic disease medical atlas to obtain the user's chronic disease index data; adjust the user's chronic disease index based on the user's lifestyle data and the user's chronic disease index data to obtain the user's chronic disease index adjustment data.

[0092] Specifically, the pre-defined chronic disease medical atlas is a model built based on extensive medical data and research findings, capable of demonstrating the correlation between different physical signs and chronic disease risks. Users' physical sign data is input into this atlas for fitting to calculate their corresponding chronic disease index. The chronic disease index represents the user's risk of developing a certain chronic disease, typically expressed as a score. Lifestyle has a direct impact on the chronic disease index. The user's lifestyle data is comprehensively analyzed in conjunction with previous chronic disease indices. For example, a healthy lifestyle reduces the risk of certain chronic diseases, while unhealthy habits increase the risk. Through these adjustments, a more accurate "adjusted chronic disease index data" is generated.

[0093] Output chronic disease index adjustment data: This data reflects the user's chronic disease risk status under current physical signs and lifestyle, and has been adjusted according to lifestyle habits and behavioral patterns.

[0094] Step S3: Obtain the user location data corresponding to the user's vital signs data, and divide the community according to the user location data and the user's chronic disease index adjustment data to obtain the user's chronic disease index community division data.

[0095] Specifically, the user's geographical location is determined using mobile phone GPS, home address, or other location identification methods. To protect privacy, this data is encrypted. Using the user's geographical location data, users living in similar areas are clustered according to their chronic disease index data. For example, in areas with a high incidence of chronic diseases, users in these areas are classified as high-risk communities, while in other areas they are classified as medium- or low-risk communities. This generates "User Chronic Disease Index Community Classification Data" reflecting each region, including the average chronic disease index and risk level of users in each community.

[0096] Step S4: Construct follow-up pathways based on the community segmentation data of users' chronic disease index to obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up work.

[0097] Specifically, based on the chronic disease index classification of different communities, the priority and frequency of follow-up visits are determined. Users in high-risk communities may require more frequent follow-up visits, while those in low-risk communities can have fewer visits. Considering geographical convenience, the pathway also needs to ensure that nurses or doctors can efficiently visit multiple patients. Based on the constructed pathway, a detailed follow-up schedule is developed, including the follow-up time, content (such as retesting of vital signs, lifestyle guidance, etc.), and required resources (medical equipment, medications, educational materials, etc.) for each user or community. The resulting "Community Chronic Disease Follow-up Plan Data" is a comprehensive follow-up scheme that clearly defines the target population, frequency, methods, and pathways for follow-up visits.

[0098] This invention, by acquiring users' vital sign and lifestyle data, allows the system to adjust nursing plans based on each user's specific chronic disease index. This method enables real-time adjustments based on users' chronic disease index and lifestyle data. Changes in lifestyle and vital signs directly affect the calculation of the chronic disease index, and adjustments to the chronic disease index update follow-up pathways and plans. By combining user location data with chronic disease index adjustment data, this method delineates community units with high and low incidence of chronic diseases, enabling better allocation of community resources, concentrating efforts on caring for high-risk patients, optimizing the use of community medical resources, and improving the efficiency of medical resources. Through precise community delineation and rational arrangement of follow-up pathways, this method can effectively reduce medical costs. High-risk patients will receive priority medical care, while the follow-up frequency for low-risk patients can be appropriately reduced, decreasing unnecessary resource consumption and reducing the overall medical cost burden.

[0099] Preferably, step S1 specifically includes:

[0100] Step S11: Collect user vital signs data and user movement data through a portable device, wherein the user vital signs data includes user heart rate data and user blood pressure data;

[0101] Specifically, wearable devices (such as smartwatches and fitness trackers) collect users' heart rate and blood pressure data in real time. These devices are typically equipped with sensors that can monitor physiological data during exercise and at rest. They also record exercise data, such as steps, type of exercise (running, cycling, etc.), and duration. The collected vital signs and exercise data are transmitted to a medical data platform via Bluetooth or Wi-Fi, where the platform performs initial cleaning and standardization to ensure a consistent format.

[0102] Step S12: Extract user lifestyle features based on user heart rate data and user exercise data to obtain user lifestyle feature data;

[0103] Specifically, by analyzing users' heart rate changes during exercise and non-exercise, lifestyle characteristics can be extracted. For example, the recovery speed of heart rate after exercise, the intensity and frequency of exercise, and resting heart rate can all serve as important indicators for assessing users' lifestyle habits (such as exercise regularity, exercise intensity, and heart health). Based on the combination of heart rate and exercise data, user lifestyle characteristic data is obtained, specifically including information such as exercise habits (regularity and intensity) and physical health status (cardiac function).

[0104] Step S13: Obtain historical user lifestyle data and historical user vital sign data;

[0105] Specifically, historical user lifestyle and vital sign data are retrieved from the database. Historical data typically includes long-term records of heart rate, blood pressure, exercise data, and lifestyle habits (such as diet and sleep). This historical data is then organized and categorized to ensure consistency with the current user's vital sign and lifestyle data structure, facilitating subsequent clustering and analysis.

[0106] Step S14: Perform lifestyle clustering calculations on historical user vital sign data based on historical user lifestyle data to obtain historical user lifestyle clustering feature data;

[0107] Specifically, clustering algorithms are used to categorize historical users' lifestyle and physical characteristics data. For example, users can be grouped into different groups based on similar exercise frequency, intensity, and physical characteristics (such as blood pressure and heart rate). This process identifies historical user groups with lifestyles similar to current users and generates multiple clustering features. Each cluster contains similar lifestyle and physical characteristics, and the generated historical user lifestyle clustering feature data reflects the patterns of how different populations' lifestyles affect their health.

[0108] Step S15: Perform fitting mapping based on user lifestyle feature data and historical user lifestyle clustering feature data to obtain user lifestyle feature fitting mapping data;

[0109] Specifically, the current user's lifestyle characteristics are matched with historical user clustering features to identify the most similar lifestyle and health patterns. This process can be achieved through feature matching, finding the historical data group most closely related to the user's current state. The fitted mapping data can reflect the similarity between the user's current lifestyle characteristics and certain specific chronic disease risk groups, thus more accurately assessing their health status.

[0110] Step S16: Obtain the user's exercise time data corresponding to the user's exercise data;

[0111] Specifically, specific exercise time data is extracted from the user's exercise data. This includes the start time, duration, and frequency of each exercise session. Exercise time data can be used to further analyze the user's exercise habits and patterns. This time data is timestamped, making it easier to combine with other feature data in subsequent evaluation and screening steps.

[0112] Step S17: Evaluate and filter the user lifestyle characteristics fitting mapping data based on the user exercise time data to obtain user lifestyle data.

[0113] Specifically, by analyzing users' exercise time data, the consistency between this data and the mapping data of their lifestyle characteristics is evaluated. For example, whether users exercise regularly according to the lifestyles of historically similar groups, whether the frequency is sufficient, and whether the intensity is reasonable. Parts that do not match the mapping data of users' lifestyle characteristics are identified, marked, and adjusted to ensure the accuracy of the user's lifestyle data. After evaluation and filtering, current lifestyle data of users is generated, which more accurately reflects users' exercise habits and their impact on health.

[0114] This invention extracts lifestyle characteristics based on user heart rate and exercise data, revealing behavioral features such as activity patterns, exercise frequency, and intensity. Lifestyle clustering features are obtained through cluster analysis of historical user lifestyle and vital sign data. Users continuously and automatically collect vital sign data, including key physiological indicators such as heart rate and blood pressure, and exercise data (such as exercise type, intensity, and steps), using portable devices (e.g., smart bracelets, smartwatches). By analyzing heart rate and exercise data, the system can identify user exercise intensity, type, and habits, accurately reflecting their true lifestyle. The system retrieves lifestyle and vital sign data from historical databases of related or similar user groups as a benchmark to establish a comparative model. Clustering calculations are performed on historical user lifestyle data to form multiple lifestyle types (e.g., active, sedentary, balanced), and features of each lifestyle type are extracted. The user's lifestyle feature data is fitted with historical clustering results to generate user lifestyle feature fitting mapping data. Through matching analysis, the user's lifestyle type is identified. The system uses users' exercise time data to fit and map their lifestyle characteristics, then further filters and evaluates the data to generate accurate lifestyle data.

[0115] Preferably, the lifestyle clustering calculation specifically involves:

[0116] Based on historical user lifestyle data, spectral clustering calculations are performed on historical user vital sign data to obtain the first historical user lifestyle cluster feature data.

[0117] Specifically, firstly, a similarity matrix is ​​constructed using historical users' lifestyle data (such as exercise frequency, intensity, and dietary habits) and vital sign data (such as heart rate and blood pressure) to reflect the degree of similarity in lifestyle and vital sign characteristics among each user. Each element in the matrix represents the similarity between two users. Dimensionality reduction methods are used to transform the similarity matrix into a lower-dimensional feature space, facilitating subsequent clustering processing. This simplifies the complex high-dimensional data into a manageable geometric structure. In the dimensionality-reduced feature space, spectral clustering methods are used to cluster users into different categories based on their lifestyle and vital sign characteristics. Users in each category have similar health and lifestyle patterns. The first cluster feature data is output, where each cluster represents a group of users with similar lifestyles and vital sign characteristics.

[0118] Based on historical user lifestyle data, graph clustering calculations are performed on historical user vital sign data to obtain second historical user lifestyle clustering feature data.

[0119] Specifically, historical user lifestyle and physical characteristic data are represented as a graph structure. Nodes in the graph represent each user, and the edge weights between nodes represent the similarity in lifestyle and physical characteristic data among users, similar to the similarity matrix used in spectral clustering. By analyzing the graph structure, highly interconnected subgroups are identified. That is, users with similar lifestyles form a tightly connected subgroup (cluster), while users with different lifestyles are assigned to other clusters. The graph clustering results are output as a second set of clustering feature data, where each cluster represents a group of users highly similar in lifestyle and physical characteristic features.

[0120] The first historical user lifestyle cluster feature data and the second historical user lifestyle cluster feature data are mapped and aligned to obtain historical user lifestyle feature aligned data.

[0121] Specifically, the first set of historical user lifestyle clustering feature data is mapped to the second set of historical user lifestyle clustering feature data, that is, the clustering results obtained by the two different methods are correlated. The core of the mapping is to find the correspondence between the same user or similar user groups under the two different clustering methods. Based on the similarity of users' lifestyle features, the results of the two clustering methods are aligned to ensure that the same or similar user groups can be correlated in the two results. This process involves feature matching and similarity measurement to ensure that the two different clustering methods can unify their output to a certain extent.

[0122] Similarity calculations and weighted merging are performed on historical user lifestyle feature alignment data to obtain historical user lifestyle cluster feature data.

[0123] Specifically, for each aligned user or user group, its similarity between the two clustering results is calculated. This process is typically based on the degree of matching between lifestyle and physical characteristics, such as the extent of differences in lifestyle features. The results of spectral and graph clustering are then weighted and merged. The weighting can be based on the accuracy of different methods or the priority of data features. For example, if spectral clustering fits physical characteristic data better, it can be given slightly higher weights; if graph clustering performs better on lifestyle data, it can be assigned more weights. The system outputs historical user lifestyle clustering feature data, generating a more accurate and comprehensive user lifestyle clustering result based on multiple clustering methods.

[0124] This invention utilizes different algorithms to cluster historical user data. Spectral clustering is suitable for handling complex relationships in high-dimensional data, while graph clustering is better suited for handling network data with complex structures. Aligning the feature data generated by different clustering algorithms ensures that features from different methods can be compared and integrated under the same criteria. Similarity calculations using the aligned data identify common characteristics and key behavioral patterns among historical users. Weighted merging further integrates the advantages of different algorithms, improving the representativeness and accuracy of the feature data. Through clustering calculations and data merging, this method enables in-depth analysis of the lifestyle characteristics of historical users, providing a more detailed assessment. Employing different clustering methods (spectral clustering and graph clustering) can mitigate the bias that may arise from a single algorithm. By weighted merging of the clustering results, the advantages of different methods can be effectively integrated, reducing potential biases caused by different algorithm choices and improving the overall accuracy of data analysis.

[0125] Preferably, the spectral clustering calculation specifically involves:

[0126] A graph is constructed based on historical user lifestyle data and historical user physical characteristics data to obtain historical user lifestyle and physical characteristics association graph data.

[0127] Specifically, relevant features are extracted from historical user lifestyle data (such as exercise frequency and dietary habits) and vital sign data (such as heart rate and blood pressure). The data is then standardized and normalized. An undirected graph is constructed from the user lifestyle and vital sign data, where each node represents a user, and the edges between nodes represent the similarity between users. Similarity can be measured based on differences in lifestyle and vital sign data, such as Euclidean distance or cosine similarity. The edge weights represent the degree of similarity between users in their lifestyle and vital sign characteristics.

[0128] Based on the historical user lifestyle characteristics correlation map data, similarity map feature encoding is performed to obtain historical user lifestyle characteristics map feature data;

[0129] Specifically, the lifestyle-related graph is processed to encode its structure into a similarity matrix. Each element of this matrix represents the similarity between two nodes (users), with rows and columns representing different users. Different weights are assigned based on the similarity between nodes. The encoding considers the similarity features of users in terms of lifestyle and physical characteristics, including normalizing the edge weights so that the matrix values ​​are between 0 and 1, facilitating subsequent calculations.

[0130] The Laplace matrix is ​​calculated based on the historical user lifestyle characteristic data to obtain the first historical user lifestyle characteristic matrix data.

[0131] Specifically, firstly, based on the similarity matrix, the degree of each node is calculated, which is the sum of the similarities between each user and other users, forming the degree matrix. The degree matrix and the similarity matrix are then used to generate the Laplacian matrix. The Laplacian matrix reflects the structural characteristics of the graph, capturing the relationships and strengths of associations between user groups. The generated Laplacian matrix will be used for subsequent eigenvalue decomposition and embedding space generation.

[0132] Based on the first historical user lifestyle characteristic matrix data, feature value decomposition and embedding space generation are performed to obtain historical lifestyle characteristic feature embedding space data.

[0133] Specifically, eigenvalue decomposition is performed on the Laplacian matrix to obtain eigenvalues ​​and eigenvectors. Eigenvalue decomposition helps analyze the connection patterns between nodes (users) in the graph. Through decomposition, high-dimensional data can be mapped to a low-dimensional space. Based on the eigenvectors, a low-dimensional embedding space is generated. This embedding space represents the spatial location of each user in terms of lifestyle and physical characteristics, and the distance between users represents their similarity. Eigenvectors with smaller eigenvalues ​​are more important; typically, the eigenvectors corresponding to the few smallest eigenvalues ​​are selected as the embedding dimensions.

[0134] Clustering is performed on spatial data embedded with historical lifestyle characteristics to obtain the first historical user lifestyle clustering characteristic data.

[0135] Specifically, clustering methods are used to group users within a low-dimensional embedding space. In this case, the closer users are in the embedding space, the more similar their lifestyles and physical characteristics are. Therefore, clustering based on distance metrics can group similar users into the same group. The clustering results are output as the first historical user lifestyle clustering feature data, with each cluster representing a group of users with similar lifestyles and physical characteristics.

[0136] This invention constructs a correlation graph from historical users' lifestyle and vital sign data, comprehensively capturing multi-dimensional relationships between users. The encoding process preserves the graph's topology and relationships between nodes, ensuring no important information is lost during conversion. The Laplace matrix reflects the graph's global structural features, including node connectivity and overall topology. Eigenvalue decomposition maps high-dimensional lifestyle and vital sign data to a low-dimensional embedding space, reducing data dimensionality and computational complexity. In the embedding space, key features are preserved, while noise and redundant information are filtered out, improving data quality. Clustering within the embedding space more accurately groups users with similar lifestyle and vital sign characteristics. The resulting first historical user lifestyle clustering feature data helps identify commonalities and differences among different user groups, providing a basis for follow-up health management.

[0137] Preferably, the similarity graph feature encoding is specifically as follows:

[0138] Multi-scale convolution calculations are performed based on historical user lifestyle characteristic correlation graph data to obtain correlation graph feature convolution layer data;

[0139] Specifically, based on the historical user lifestyle and physical characteristic association graph, the features of each node (user) are processed through multi-scale convolution operations. The purpose of multi-scale convolution is to extract local and global features of user nodes from different scales (i.e., neighborhood ranges) to capture the similarity of their lifestyle and physical characteristics across different ranges. After convolution processing, the output association graph feature convolutional layer data is generated, containing the multi-scale features of each user in the graph. Different scales reflect the similarity between users and other users within different ranges.

[0140] Global and local multi-head self-attention calculation is performed on the convolutional layer data of the association graph features to obtain the first association graph feature self-attention data;

[0141] Specifically, a global-local multi-head self-attention mechanism is applied to the convolutional layer data of the association graph features. The self-attention mechanism allows the model to focus on the interaction between each node and other nodes. Global self-attention captures the similarity between users and distant users, while local self-attention focuses on the relationship between a user and its neighboring nodes. After global-local multi-head self-attention calculation, the first association graph feature self-attention data is obtained. This data represents the interdependencies between different user nodes, as well as the overall patterns of lifestyle and physical characteristics, in the case of multiple heads.

[0142] Perform masked multi-head self-attention calculation on the convolutional layer data of the association graph to obtain the feature self-attention data of the second association graph.

[0143] Specifically, for the convolutional layer data of the association graph, a multi-head self-attention operation with a masking mechanism is applied. Masked self-attention focuses on certain nodes or features by masking the connections of some nodes, thereby reducing unnecessary interference and emphasizing the core features of the nodes. The output is second association graph feature self-attention data, which enhances the attention between specific nodes through the masking mechanism, especially helping to distinguish nodes with low similarity or irrelevant connections.

[0144] Pooling layer calculation is performed on the self-attention data of the first association graph feature to obtain the association graph feature pooling layer data, and feature distillation is performed on the self-attention data of the second association graph feature to obtain the association graph feature distillation data.

[0145] Specifically, pooling operations are performed on the self-attention data of the first association graph feature, using either max pooling or average pooling. Pooling layers reduce the feature dimensionality, retaining the most important feature information. This helps reduce data complexity while preserving the main features relevant to similarity judgment. Feature distillation is then performed on the self-attention data of the second association graph feature. The purpose of distillation is to extract and compress the most representative features while removing unnecessary redundant information. Through this method, the self-attention data of the second association graph feature is simplified into the core similarity representation.

[0146] Assuming there are 1000 users, each user's features are represented as a 128-dimensional vector. Using weight analysis or feature selection methods, such as calculating the variance and mean of each feature dimension, or using attention weights generated by the model, we determine which dimensions have the greatest impact on the results. We calculate the variance of each feature dimension. The feature dimensions are sorted from highest to lowest variance, and the top 64 dimensions with the highest variance are selected as the most important features, as high variance usually indicates high discriminative power among different users. The original 128-dimensional feature data is compressed to 64 dimensions to extract the most representative features. Linear transformations or dimensionality reduction techniques, such as Principal Component Analysis (PCA) or weight-based weighted averaging, are used to compress the high-dimensional features. After feature distillation, a 1000×64 matrix is ​​obtained, representing the compressed 64-dimensional features for each user. This process preserves the most important information from the original self-attention data while reducing computational complexity.

[0147] Based on the feature pooling layer data and feature distillation data of the association graph, a similar graph is constructed to obtain the feature data of historical user lifestyle characteristics.

[0148] Specifically, the final similarity graph is generated by combining the self-attention data of the first association graph after pooling and the self-attention data of the second association graph after distillation. Nodes in the similarity graph represent users, and edge weights represent the similarity between users. Using data from these two sources, the constructed similarity graph better reflects the relationships between users' lifestyles and physical characteristics. The resulting similarity graph includes the similarity between user nodes, reflecting the overall distribution and association patterns of lifestyles and physical characteristics.

[0149] This invention utilizes feature data with strong representational capabilities to construct more accurate similarity graphs, providing a solid foundation for subsequent clustering analysis and improving the accuracy of chronic disease risk assessment and care plan development. Through multiple attention mechanisms and feature distillation, the model can better cope with noise and missing data, improving adaptability to new data and predictive performance. Precise feature encoding and similarity graph construction enable a deeper understanding of users' lifestyles and health status, thus supporting customized care interventions and follow-up plans. Convolutional operations for graph-structured data are suitable for processing data in non-Euclidean spaces, improving the ability to model complex network relationships. Global-local multi-head self-attention can simultaneously consider the local neighborhood information and global structural information of nodes in the graph, deeply understanding the complex relationships in user data. By masking part of the input during training, the model can learn more robust feature representations, reducing overfitting. Pooling operations can simplify feature representations, retain the most important information, reduce computational complexity, and prevent overfitting. Feature distillation extracts knowledge from complex models and transfers it to simplified models, achieving model lightweighting, improving computational efficiency, and maintaining performance. By combining the features of pooling and distillation, the model can leverage feature information from different levels and sources to enhance the understanding of users' health status. Similarity graphs help to more accurately identify user groups, improve the performance of clustering algorithms, and thus enhance the effectiveness of community chronic disease care follow-up programs.

[0150] Preferably, the graph clustering calculation specifically involves:

[0151] A similarity matrix is ​​constructed based on historical user lifestyle data and historical user physical characteristics data to obtain historical user lifestyle characteristics similarity matrix data.

[0152] Specifically, numerical features are extracted from historical user lifestyle data (such as exercise and dietary habits) and vital sign data (such as heart rate and blood pressure) to ensure data standardization and normalization, enabling unified measurement of different types of data. Similarity metrics (such as Euclidean distance and cosine similarity) are used to calculate the similarity between each pair of users. Higher similarity indicates a closer similarity in lifestyle and vital signs. Each element in the similarity matrix represents the similarity between two users; larger values ​​indicate greater similarity in their lifestyles and vital signs.

[0153] Based on the historical user lifestyle characteristics similarity matrix data, an adjacency matrix and a degree matrix are constructed to obtain historical user lifestyle characteristics adjacency matrix data and historical user lifestyle characteristics degree matrix data, respectively.

[0154] Specifically, an adjacency matrix is ​​constructed based on the similarity matrix. The adjacency matrix uses 0 and 1 to represent whether there is a connection between user nodes. A similarity threshold is set; if the similarity between two users exceeds the threshold (any value between 0.6 and 0.8), the corresponding position in the adjacency matrix is ​​marked as 1, indicating that the two users are connected; otherwise, it is marked as 0. The degree matrix represents the connectivity of each node (user), specifically the total number of similar connections between each user node and other nodes. The degree matrix is ​​a diagonal matrix, with the elements on the main diagonal representing the degree of each node (i.e., the sum of the elements in each row of the adjacency matrix). The adjacency matrix reflects the adjacency relationship between users, while the degree matrix provides the connectivity degree of each user node.

[0155] The Laplace matrix is ​​calculated based on the historical user lifestyle characteristic adjacency matrix data and the historical user lifestyle characteristic degree matrix data to obtain the first historical user lifestyle characteristic matrix data.

[0156] Specifically, the Laplacian matrix of the graph is calculated using the degree matrix and the adjacency matrix. The Laplacian matrix is ​​defined as the degree matrix minus the adjacency matrix. It captures the connectivity relationships between nodes in the graph, as well as the feature differences between a node and its neighborhood. The Laplacian matrix reveals the overall structure among user groups, helping to identify local and global connections in the graph, and providing a foundation for subsequent eigenvalue decomposition and graph embedding.

[0157] Based on the first historical user lifestyle characteristic matrix data, eigenvalue decomposition and graph embedding are performed to generate historical user lifestyle characteristic embedding data.

[0158] Specifically, eigenvalue decomposition is performed on the Laplacian matrix to obtain a set of eigenvalues ​​and corresponding eigenvectors. Eigenvalue decomposition reveals the geometric structure of the graph, and by analyzing the magnitude of the eigenvalues, the distribution of different subgroups in the graph can be determined. Based on the eigenvectors, high-dimensional lifestyle characteristic data is embedded into a low-dimensional space. Each dimension of this embedding space represents an important feature of the user group, and the distance in the embedding space reflects the similarity between users; the closer the distance, the more similar the lifestyles and characteristics. The embedded data is output, and the data in the embedding space is used for subsequent clustering processing to identify different user groups in the low-dimensional space using clustering methods.

[0159] Hierarchical clustering is performed based on historical user lifestyle characteristics embedded data to obtain second historical user lifestyle clustering feature data.

[0160] Specifically, hierarchical clustering is performed on users within the embedding space. Hierarchical clustering forms a tree structure (i.e., a "tree diagram") by recursively merging similar user groups. Each level represents a different grouping of user similarity, with the bottom level representing each individual user and the top level representing the merged group of all users. Through hierarchical clustering, second-historical user lifestyle clustering feature data is obtained. This data reflects the distribution of different groups in terms of lifestyle and physical characteristics, and can be further used for health management and care follow-up.

[0161] The graph clustering method in this invention can handle nonlinear, high-dimensional, and heterogeneous data, exhibiting strong adaptability in complex user data scenarios and making it suitable for community needs analysis in precision care for chronic diseases. By calculating a similarity matrix, the similarity between users is accurately measured, helping to identify user groups with similar lifestyles and physical characteristics, thus facilitating the development of more precise care plans that conform to the general characteristics of community users. The similarity matrix can uncover implicit associations from different dimensions of lifestyle and physical characteristic data, helping to discover potential health association models between users or between users and lifestyles. The adjacency matrix constructs relationships between users through threshold filtering, and the generated graph structure reflects the global connectivity of users. The combination of the degree matrix and the adjacency matrix provides necessary topological information for subsequent Laplace matrix calculation. Eigenvalue decomposition effectively reduces dimensionality, simplifying high-dimensional complex data into a low-dimensional embedding space while retaining important feature information and avoiding information loss. The data in the embedding space can more clearly reflect the feature similarity of user groups, facilitating more accurate clustering analysis and improving clustering results. Hierarchical clustering can capture the hierarchical structure within a user group and identify user groups at different granularities. It is particularly suitable for identifying subgroups of users with different potential needs for chronic diseases.

[0162] Preferably, the evaluation and screening specifically includes:

[0163] Based on user exercise time data, predict user biological clock behavior to obtain user biological clock behavior data;

[0164] Specifically, firstly, user exercise time data is collected, including the time period, frequency, and duration of each user's daily exercise. This data allows observation of users' daily exercise patterns. Combining this exercise time data, predictions of users' circadian rhythm behavior are made. By analyzing the frequency and time distribution of users' exercise, their daily routines and circadian rhythms can be inferred. For example, people who exercise in the morning tend to go to bed early and wake up early, while those who exercise in the evening are nocturnal active. Based on the exercise time data, user circadian rhythm behavior data is generated, including peak activity periods, daytime and nighttime exercise preferences, and sleep patterns.

[0165] Based on user exercise data and user biological clock behavior data, indoor and outdoor lifestyles are filtered to obtain user indoor and outdoor lifestyle filtering data.

[0166] Specifically, user activity data is used to analyze the type of activity (e.g., running, walking, cycling) and the location (indoor or outdoor) using GPS data. Combined with user circadian rhythm data, further analysis is conducted to determine whether users' daily habits lean towards indoor or outdoor activities. By comparing activity type, time, and location, users' indoor and outdoor lifestyles are filtered. For example, if a user exercises indoors in the early morning or at night, they may have an indoor lifestyle; if they frequently engage in outdoor activities during the day, they may have an outdoor lifestyle. This generates user lifestyle filtering data, reflecting user lifestyle preferences (e.g., indoor / outdoor bias) and how these preferences interact with users' circadian rhythm behaviors.

[0167] User lifestyle data is obtained by fitting data based on user internal and external lifestyle filtering data and user lifestyle feature fitting mapping data.

[0168] Specifically, the user's internal and external lifestyle screening data is matched with previously generated user lifestyle feature fitting mapping data. The purpose of fitting is to compare the user's lifestyle (including their circadian rhythm behavior, exercise habits, lifestyle preferences, etc.) with lifestyle patterns in historical data. During the fitting process, similar patterns between the user's current lifestyle and historical lifestyle features are identified, ensuring that the user's lifestyle habits are closely integrated with existing chronic disease risk assessment data. Through this matching, the understanding of the user's lifestyle can be adjusted and improved. This generates user lifestyle data that not only reflects the user's lifestyle but also, through fitting mapping, obtains a precise lifestyle profile based on historical feature data, providing a basis for subsequent health management and care follow-up.

[0169] In this invention, a portable device collects users' exercise time data. By analyzing the user's daily exercise start and end times, frequency, and duration, the system can accurately infer the user's daily routine based on this data. Combining user exercise data (such as GPS location, step count, indoor / outdoor detection data, etc.) with circadian rhythm data, the system can determine whether the user's exercise takes place outdoors or whether they spend most of their time indoors. The system identifies the ratio of indoor to outdoor activities: if most exercise is outdoors (e.g., walking, running), it is categorized as "primarily outdoor activity"; if most exercise occurs indoors (e.g., walking at home, working out in a gym), it is categorized as "primarily indoor activity." The system fits and maps the user's current indoor / outdoor lifestyle data with historical lifestyle data. By matching similar historical data, it predicts the user's future lifestyle trends and outputs lifestyle data. If the user's lifestyle highly matches the behavioral patterns of historically active users, the system classifies the user as active; if the user exhibits excessive indoor activity and a lack of outdoor activity, the system identifies the user as sedentary. The system maps users’ actual exercise data with historical users’ lifestyle characteristics data, matching similar patterns and features to obtain accurate user behavior data. Compared with interface input, it can better reflect the actual situation and provides an accurate data foundation for the development of follow-up plans.

[0170] Preferably, step S3 specifically includes:

[0171] Step S31: Obtain the user location data corresponding to the user's vital signs data;

[0172] Specifically, GPS data is obtained from user devices (such as mobile phones and smart bracelets), or static information such as home addresses is used to determine the user's geographical location. Location data can be latitude and longitude coordinates, or the user's location can be marked by postal codes, etc. The collected user vital signs data is correlated with their location data to ensure that each user's health status corresponds to their geographical location. Vital signs data include chronic disease-related indicators such as heart rate and blood pressure, while location data is used to identify the user's geographical environment.

[0173] Step S32: Correlate user location data and user chronic disease index adjustment data to obtain user chronic disease index location correlation data;

[0174] Specifically, user chronic disease index adjustment data (such as chronic disease risk scores) is integrated with user location data to form correlation data between location and chronic disease index. User location data (such as home address, latitude and longitude) can be used to mark the distribution of the user in a certain geographical area. Based on the user's chronic disease index and location data, the distribution of the user's chronic disease index in different areas is analyzed to obtain user chronic disease index location correlation data.

[0175] Step S33: Perform regional clustering calculations based on the location association data of the user's chronic disease index to obtain the location clustering feature data of the user's chronic disease index;

[0176] Specifically, regional clustering calculations are performed using location-related data of users' chronic disease index. The purpose of clustering is to group users with similar chronic disease index levels into the same region. The clustering method groups adjacent users with similar chronic disease indices together, forming different health risk areas, based on their geographical location and chronic disease index. The clustering result is location-based clustering feature data of users' chronic disease indices, representing the distribution of users' chronic disease indices across different regions. Each cluster represents a region containing a group of users with similar chronic disease risks.

[0177] Step S34: Obtain community medical resource data;

[0178] Specifically, medical resource data is collected from community health centers, hospitals, and other medical institutions. This data includes the number of hospital beds, doctors and nurses, the distribution of medical equipment, and the coverage of community health services; alternatively, it includes the allocation of follow-up medical personnel, follow-up medical resources, and follow-up distance limitations. This medical resource data is standardized to ensure its usability for subsequent matching and segmentation. Standardized data includes the medical service capabilities and resource allocation levels of different communities.

[0179] Step S35: Match and divide the data based on community medical resources data and user chronic disease index location clustering feature data to obtain user chronic disease index community division data.

[0180] Specifically, user chronic disease index location clustering feature data is matched with community medical resource data. The goal of this matching is to ensure that medical resources are effectively allocated to areas with a high incidence of chronic diseases. For example, high-risk areas for chronic diseases should be prioritized for matching with communities rich in medical resources to meet high demand. Through this matching process, user chronic disease index community segmentation data is generated. This data divides users into zones based on their chronic disease risk and available medical resources in their respective areas, ensuring that each community receives appropriate medical service support according to the user's health needs.

[0181] This invention combines users' vital sign data with geographic location information to deeply analyze the distribution of users' chronic disease risks, helping medical institutions optimize the allocation of community nursing resources. Through regional clustering and medical resource matching, medical resources can be allocated efficiently and rationally within the community, ensuring that high-risk users receive priority care and improving the efficiency of community chronic disease management. By rationally matching medical resources with users' health conditions, this method can promote the balanced distribution of medical resources within the community, avoiding problems of resource concentration or uneven distribution. Through precise community segmentation, follow-up pathways can be optimized according to regional divisions, improving the efficiency of follow-up work and reducing follow-up costs. The community segmentation and resource matching steps enable medical institutions to flexibly adjust nursing strategies according to the health needs of users in each region, ensuring that all users receive nursing services suitable for their health conditions.

[0182] Preferably, step S4 specifically includes:

[0183] Step S41: Construct visitation routes based on user chronic disease index community segmentation data to obtain primary visitation route data;

[0184] Specifically, primary visit routes are constructed using community segmentation data based on users' chronic disease index. This can be optimized using a Geographic Information System (GIS) based on users' geographic location and community distribution, ensuring that the visit routes for healthcare personnel are optimal in terms of time and distance. Primary visit routes should cover all users at high chronic disease risk within the community. Geographic proximity is prioritized, enabling follow-up personnel to efficiently visit multiple users and reduce unnecessary repetitive routes. The generated primary visit route data includes the specific communities and user locations that follow-up personnel need to visit.

[0185] Step S42: Prioritize the community division data based on the user's chronic disease index community division data to obtain community division priority data;

[0186] Specifically, communities are prioritized based on user chronic disease index data. Communities with higher chronic disease indices should receive higher priority, meaning these areas require more frequent and timely follow-up and care. Prioritization considers the proportion of high-risk users within the community, the overall chronic disease index level, and available medical resources. These data are weighted to determine the follow-up priority of each community. The generated community prioritization data sorts each community by risk level, facilitating subsequent matching of follow-up resources.

[0187] Step S43: Obtain follow-up personnel data;

[0188] Specifically, data on follow-up personnel involved in community chronic disease care is collected, including the number, professional skills, working hours, and geographical location of doctors, nurses, health consultants, etc. This data is then standardized and categorized according to their professional fields, service capabilities, and working hours to provide input for subsequent follow-up resource forecasting. The output of follow-up personnel data ensures an effective match between their information and the community's needs.

[0189] Step S44: Estimate follow-up resources based on follow-up personnel data to obtain follow-up resource data;

[0190] Specifically, based on the data from follow-up personnel, their service capabilities, working hours, and coverage areas are assessed. Combining the number of users and risk levels within the community, the workload and resource allocation requirements for each follow-up personnel are estimated. Based on the number of users each follow-up personnel can handle and the service frequency, the total follow-up resources required for the entire community are projected. This includes human resource allocation, time management, and transportation needs. The generated follow-up resource data includes the allocation of follow-up personnel, required equipment, service frequency, and the projected overall resource requirements.

[0191] Step S45: Perform priority matching based on follow-up resource data and community division priority data to obtain community division follow-up matching data;

[0192] Specifically, follow-up resource data is matched with community-level priority data. High-priority communities should be allocated more follow-up personnel and medical resources, while low-priority communities can have fewer follow-up visits. Dynamic allocation is based on the community's chronic disease risk level and available follow-up personnel resources to ensure maximum resource utilization. Simultaneously, the matching degree between the follow-up personnel's professional skills and the community's needs is considered to ensure accurate resource allocation. The matching results constitute the community-level follow-up matching data, determining which follow-up personnel are responsible for which community, as well as the follow-up frequency and resource allocation.

[0193] Step S46: Map the primary visit route data to the community-based follow-up matching data to obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up work.

[0194] Specifically, the community-based follow-up matching data is mapped to the initial visit route data, and the visit routes are re-optimized. Taking into account the specific allocation of follow-up personnel, community priorities, and time arrangements, the initial routes are adjusted to match the allocation of follow-up resources. The optimized visit routes ensure that follow-up personnel can visit communities according to their priority, while minimizing route overlap and wasted time. The generated community chronic disease follow-up plan data includes visit routes, follow-up personnel allocation, service frequency, and follow-up plans for each community, used to guide follow-up operations.

[0195] Specifically, in another embodiment, community-based follow-up matching data is performed on the initial visit route data to obtain visit route point data; a follow-up plan is generated based on the visit route point data to obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up operations.

[0196] The initial visit route data is generated based on the community chronic disease index and user location, representing the communities and corresponding user groups that follow-up personnel need to visit. The initial version of the route primarily determines the optimal route based on geographical location and chronic disease risk, without considering resource and priority matching. Community-based follow-up matching data includes the priority of each community (based on the chronic disease index) and the specific resources allocated to each follow-up personnel. This data influences the final follow-up route, ensuring that high-priority communities receive timely follow-ups and optimizing the route based on the follow-up personnel's resource allocation capabilities. The initial visit route is combined with the community-based follow-up matching data. High-priority communities will be prioritized for inclusion in the final route, and the priority and resource allocation of each point along the route will be reassessed. If some low-priority communities cannot be covered in the first round of follow-ups, their arrangements can be postponed, while high-priority communities will receive priority visits. After matching, visit route point data is generated. This data represents the user location points that follow-up personnel need to visit in actual operation, based on the community's chronic disease risk and resource allocation. These location points cover high-priority communities and reasonable resource allocation routes.

[0197] The data on the visit routes determines each point that the follow-up personnel actually need to pass through, ensuring that resources and priorities are allocated reasonably. Each point can represent the location of a specific community or user group, and is determined based on factors such as geographical location, chronic disease index, and resource needs.

[0198] Based on the data from these visit routes, a detailed follow-up plan is generated, including: setting specific time schedules for each visit point to ensure efficient coverage by follow-up personnel; placing high-priority points in the first half of the visit route to ensure early coverage; developing personalized follow-up content for each user based on their chronic disease index and historical health data. For example, high-risk users may require blood pressure and blood sugar monitoring and medication guidance, while low-risk users may only need health advice and lifestyle guidance; assigning appropriate follow-up personnel to each visit point to ensure their skills match the user's needs (e.g., nurses, doctors); determining the medical equipment and resources to be carried during the follow-up, such as blood pressure monitors, blood glucose meters, and medications. Material preparation is planned according to the user needs at each visit point to ensure that the resources carried match the visit content; using Geographic Information System (GIS) technology to assist in route planning and optimizing the visit sequence based on the follow-up personnel's schedules; and calculating factors such as transportation routes and time costs to make the follow-up plan more efficient. The follow-up plan data includes the specific visit sequence, visit time, supplies to be brought, and specific follow-up services required by each follow-up personnel.

[0199] After generating the follow-up plan, follow-up personnel begin carrying out community chronic disease care follow-up work. According to the generated plan, follow-up personnel visit each location to provide health management services to users. During the follow-up, personnel record users' health data in real time (such as vital sign monitoring, medication use, etc.). This data can be uploaded to a cloud platform via mobile devices to ensure all data is updated in real time. If new health risks are discovered during the follow-up (such as abnormally high blood pressure), subsequent follow-up plans can be adjusted accordingly, and further medical interventions can be prioritized. Follow-up routes and priorities can be adjusted as needed based on the follow-up results. If certain users require emergency treatment or other medical services, they can be prioritized for the next round of follow-ups. By accumulating users' health data through multiple follow-ups, follow-up personnel can gradually adjust users' chronic disease management plans to achieve precision care. For example, improvements or deteriorations in a user's chronic disease index can be reflected in the follow-up plan, dynamically adjusting service content and frequency.

[0200] This invention ensures the rational use of follow-up resources, improves follow-up efficiency, and reduces resource waste through preliminary path planning and prioritization. Based on users' chronic disease indices and community classifications, high-risk patients are prioritized to maximize follow-up effectiveness and avoid unnecessary visits. Combining user health status, community priority, and follow-up personnel capabilities, the most suitable follow-up plan is provided for each community, ensuring basic health management effectiveness while minimizing medical costs. The follow-up plan can be dynamically adjusted based on changes in resources, personnel, and patient health status, ensuring flexibility in follow-up work.

[0201] Preferably, this application also provides a community-based precision care follow-up system for chronic diseases, used to perform the community-based precision care follow-up method for chronic diseases as described above. The community-based precision care follow-up system for chronic diseases includes:

[0202] The user data collection module is used to acquire user vital sign data and user lifestyle data.

[0203] The user chronic disease index adjustment module is used to fit user vital sign data and a preset chronic disease medical atlas to obtain user chronic disease index data; and to adjust user chronic disease index based on user lifestyle data and user chronic disease index data to obtain user chronic disease index adjustment data.

[0204] The user chronic disease index community division module is used to obtain user location data corresponding to user vital sign data, and to divide the community based on user location data and user chronic disease index adjustment data to obtain user chronic disease index community division data.

[0205] The community chronic disease nursing follow-up pathway construction module is used to construct follow-up pathways based on the user's chronic disease index community division data, and obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up operations.

[0206] The beneficial effects of this invention are as follows: By fitting user vital sign data with a chronic disease medical atlas, the chronic disease risk of each patient can be more accurately assessed, forming a dynamic chronic disease index that reflects the trend of individual health status changes over time. The chronic disease index is adjusted in real time according to the user's lifestyle, ensuring that the care plan adapts to changes in the patient's actual health condition, thus making the care plan more flexible and dynamic, and effectively responding to sudden changes in the patient's lifestyle or health status. Through community segmentation, patient groups with similar health risks are identified and clustered, forming community health zones with different risk levels. This space-based chronic disease management method can effectively improve the allocation efficiency of community medical resources. By combining community priority segmentation, follow-up resource data, and follow-up personnel capabilities, the optimal follow-up path is dynamically generated, thereby significantly improving the efficiency and effectiveness of follow-up. This invention not only enables precise health management for individuals but also achieves broader community health management goals through cluster analysis and regional segmentation.

[0207] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended application documents rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of the equivalents of the application documents be incorporated into the invention.

[0208] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for precise nursing follow-up of chronic diseases in the community, characterized in that, Includes the following steps: Step S1: Obtain user vital signs data and user lifestyle data; Step S2: Fit the user's vital signs data and the preset chronic disease medical atlas to obtain the user's chronic disease index data; adjust the user's chronic disease index based on the user's lifestyle data and the user's chronic disease index data to obtain the user's chronic disease index adjustment data. Step S3: Obtain the user location data corresponding to the user's vital signs data, and divide the community according to the user location data and the user's chronic disease index adjustment data to obtain the user's chronic disease index community division data. Step S4: Construct follow-up pathways based on user chronic disease index community segmentation data to obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up work; Step S1 specifically involves: Step S11: Collect user vital signs data and user movement data through a portable device, wherein the user vital signs data includes user heart rate data and user blood pressure data; Step S12: Extract user lifestyle features based on user heart rate data and user exercise data to obtain user lifestyle feature data; Step S13: Obtain historical user lifestyle data and historical user vital sign data; Step S14: Perform lifestyle clustering calculations on historical user vital sign data based on historical user lifestyle data to obtain historical user lifestyle clustering feature data; Step S15: Perform fitting mapping based on user lifestyle feature data and historical user lifestyle clustering feature data to obtain user lifestyle feature fitting mapping data; Step S16: Obtain the user's exercise time data corresponding to the user's exercise data; Step S17: Evaluate and filter the user lifestyle characteristics fitting mapping data based on the user exercise time data to obtain user lifestyle data; The lifestyle clustering calculation specifically involves: Based on historical user lifestyle data, spectral clustering calculations are performed on historical user vital sign data to obtain the first historical user lifestyle cluster feature data. Based on historical user lifestyle data, graph clustering calculations are performed on historical user vital sign data to obtain second historical user lifestyle clustering feature data. The first historical user lifestyle cluster feature data and the second historical user lifestyle cluster feature data are mapped and aligned to obtain historical user lifestyle feature aligned data. Similarity calculations and weighted merging are performed on historical user lifestyle feature alignment data to obtain historical user lifestyle cluster feature data; The spectral clustering calculation specifically involves: A graph is constructed based on historical user lifestyle data and historical user physical characteristics data to obtain historical user lifestyle and physical characteristics association graph data. Based on the historical user lifestyle characteristics correlation map data, similarity map feature encoding is performed to obtain historical user lifestyle characteristics map feature data; The Laplace matrix is ​​calculated based on the historical user lifestyle characteristic data to obtain the first historical user lifestyle characteristic matrix data. Based on the first historical user lifestyle characteristic matrix data, feature value decomposition and embedding space generation are performed to obtain historical lifestyle characteristic feature embedding space data. Based on the historical lifestyle characteristics, spatial data is embedded and clustered to obtain the first historical user lifestyle cluster characteristic data; The specific encoding of similarity graph features is as follows: Multi-scale convolution calculations are performed based on historical user lifestyle characteristic correlation graph data to obtain correlation graph feature convolution layer data; Global and local multi-head self-attention calculation is performed on the convolutional layer data of the association graph features to obtain the first association graph feature self-attention data; Perform masked multi-head self-attention calculation on the convolutional layer data of the association graph to obtain the feature self-attention data of the second association graph. Pooling layer calculation is performed on the self-attention data of the first association graph feature to obtain the association graph feature pooling layer data, and feature distillation is performed on the self-attention data of the second association graph feature to obtain the association graph feature distillation data. Based on the feature pooling layer data and feature distillation data of the association graph, a similar graph is constructed to obtain the feature data of historical user lifestyle characteristics. The graph clustering calculation specifically involves: A similarity matrix is ​​constructed based on historical user lifestyle data and historical user physical characteristics data to obtain historical user lifestyle characteristics similarity matrix data. Based on the historical user lifestyle characteristics similarity matrix data, an adjacency matrix and a degree matrix are constructed to obtain historical user lifestyle characteristics adjacency matrix data and historical user lifestyle characteristics degree matrix data, respectively. The Laplace matrix is ​​calculated based on the historical user lifestyle characteristic adjacency matrix data and the historical user lifestyle characteristic degree matrix data to obtain the first historical user lifestyle characteristic matrix data. Based on the first historical user lifestyle characteristic matrix data, feature value decomposition and graph embedding are performed to generate historical user lifestyle characteristic embedding data. Hierarchical clustering is performed based on historical user lifestyle characteristics embedded data to obtain second historical user lifestyle clustering feature data. The evaluation and screening process specifically includes: Based on user exercise time data, predict user biological clock behavior to obtain user biological clock behavior data; Based on user exercise data and user biological clock behavior data, indoor and outdoor lifestyles are filtered to obtain user indoor and outdoor lifestyle filtering data. Based on the user's internal and external lifestyle filtering data and the user's lifestyle characteristics fitting mapping data, the user's lifestyle data is obtained by fitting the data. Step S3 is as follows: Step S31: Obtain the user location data corresponding to the user's vital signs data; Step S32: Correlate user location data and user chronic disease index adjustment data to obtain user chronic disease index location correlation data; Step S33: Perform regional clustering calculations based on the location association data of the user's chronic disease index to obtain the location clustering feature data of the user's chronic disease index; Step S34: Obtain community medical resource data; Step S35: Match and divide the data based on community medical resource data and user chronic disease index location clustering feature data to obtain user chronic disease index community division data; Step S4 is as follows: Step S41: Construct visitation routes based on user chronic disease index community segmentation data to obtain primary visitation route data; Step S42: Prioritize the community division data based on the user's chronic disease index to obtain community division priority data; Step S43: Obtain follow-up personnel data; Step S44: Estimate follow-up resources based on follow-up personnel data to obtain follow-up resource data; Step S45: Perform priority matching based on follow-up resource data and community division priority data to obtain community division follow-up matching data; Step S46: Map the primary visit route data to the community-based follow-up matching data to obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up work.

2. A community-based precision nursing follow-up system for chronic diseases, characterized in that, For performing the community-based precision care follow-up method for chronic diseases as described in claim 1, the community-based precision care follow-up system for chronic diseases includes: The user data collection module is used to acquire user vital sign data and user lifestyle data. The user chronic disease index adjustment module is used to fit user vital sign data and a preset chronic disease medical atlas to obtain user chronic disease index data; and to adjust user chronic disease index based on user lifestyle data and user chronic disease index data to obtain user chronic disease index adjustment data. The user chronic disease index community division module is used to obtain user location data corresponding to user vital sign data, and to divide the community based on user location data and user chronic disease index adjustment data to obtain user chronic disease index community division data. The community chronic disease nursing follow-up pathway construction module is used to construct follow-up pathways based on the user's chronic disease index community division data, and obtain community chronic disease follow-up plan data for community chronic disease nursing follow-up operations.

Images