A knowledge and data-integrated chronic disease collaborative management method and system

Abstract

The present invention discloses a method and system for collaborative management of chronic diseases by integrating knowledge and data, and relates to the field of medical management technology. It includes generating a feature data set, processing the feature data set using a feature gain modeling method, and generating an enhanced feature data set; dynamically generating a comprehensive knowledge graph for chronic diseases, supplementing nodes and optimizing weights of the knowledge graph, and generating a patient-specific knowledge sub-graph; using the specific knowledge sub-graph to perform multi-time series predictions, and perform short-term and long-term trend predictions; combining the prediction results, and constructing a patient health trajectory model through state transition probabilities; based on the dynamic risk labeling results of the health trajectory model, combined with the specific knowledge sub-graph, comprehensively evaluating the patient's health status, and generating personalized management recommendations based on the evaluation results. The collaborative management platform of the present invention can promote cooperation between different medical institutions and medical staff, realize information sharing and resource integration, and improve the overall level of medical services.

Description

Technical Field

[0001] The present invention relates to the field of medical management technology, and in particular to a method and system for collaborative management of chronic diseases by integrating knowledge and data. Background Art

[0002] In the field of chronic disease management, with the aging of the population and changes in lifestyle, multi-disease co-occurrence has become an important issue. In order to meet this challenge, a variety of chronic disease collaborative management methods have emerged. These methods include community-based chronic disease health management support centers, which realize the exchange and sharing of patient health information at all levels of medical and health institutions by integrating technologies such as the Internet, APP, and big data cloud computing. In addition, there are also individualized chronic disease management and diagnosis and treatment platforms based on mobile Internet. These platforms have the characteristics of mobility, personalization and precision. By collecting data on basic information, physical signs, daily exercise, health examinations, etc. of patients, and using risk assessment models formed by big data algorithms, they provide targeted individualized health management and intervention plans for chronic disease patients. There is also a chronic disease collaborative management system for multiple diseases. The system includes a client terminal and a service engine. Through the path-driven chronic disease collaborative management model ontology, the monitoring data transmitted by the patient terminal is inferred, and full-scenario decision support information for multiple chronic diseases is obtained and sent to the doctor's work platform.

[0003] Although the existing collaborative management methods for chronic diseases have improved the management efficiency and quality of chronic diseases to a certain extent, there are still some shortcomings. First, the existing systems rely more on information technology, but lack an effective mechanism to deeply integrate clinical experience, medical knowledge and patient data to provide more accurate individualized treatment plans and management decision support. Secondly, the existing chronic disease management system has the phenomenon of unsatisfactory management effect or conflicting treatment plans when multiple diseases coexist, because the existing chronic disease management guidelines often only focus on the management of a single type of disease, and do not refine the commonality and specificity of the management and treatment plans for different diseases. Finally, despite the application of big data and artificial intelligence technologies, the integration of knowledge and data in chronic disease management is still insufficient, resulting in limited personalized and precise management services. These problems indicate that a new method and system is needed to more effectively manage chronic patients with multiple diseases, especially in terms of knowledge and data integration. Summary of the invention

[0004] In view of the limitations of existing chronic disease collaborative management technologies, a chronic disease collaborative management method and system that integrates knowledge and data is proposed in the present invention.

[0005] Therefore, the problem to be solved by the present invention is how to achieve more efficient, accurate and personalized management of multiple co-occurring chronic diseases.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides a method for collaborative management of chronic diseases by integrating knowledge and data, which includes: performing multi-dimensional hierarchical collection of patient characteristic data to generate a characteristic data set, and using a characteristic gain modeling method to process data of different dimensions in the characteristic data set to generate an enhanced characteristic data set; dynamically generating a comprehensive knowledge graph for chronic diseases by combining medical literature, clinical guidelines and patient group big data; at the same time, performing personalized node supplementation and weight optimization on the comprehensive knowledge graph to generate a patient-specific knowledge sub-graph, and performing bidirectional correlation analysis; performing multi-time series prediction using the specific knowledge sub-graph to perform short-term and long-term trend predictions on chronic disease health indicators; and combining the prediction results, constructing a patient health trajectory model through state transition probability to reflect the dynamic evolution of different health states and future risk nodes; comprehensively evaluating the patient's health state based on the dynamic risk labeling results of the health trajectory model, combined with the node weights and associated edges in the specific knowledge sub-graph, and generating personalized management recommendations based on the evaluation results.

[0008] As a preferred solution of the method for collaborative management of chronic diseases by integrating knowledge and data of the present invention, the following steps are included: personalized node supplementation and weight optimization of the comprehensive knowledge graph, generation of patient-specific knowledge sub-graph, and bidirectional association analysis: based on the enhanced feature data set Dynamically mine individual-specific features to generate supplementary nodes and associated edges. Combined with the bidirectional relationship derivation method, the association strength between the supplementary nodes and the basic nodes is calculated, and added to the comprehensive knowledge graph to generate the patient's preliminary specific knowledge sub-graph; according to the node and edge characteristics, a weight adjustment method based on the feature contribution rate is used to optimize the preliminary specific knowledge sub-graph to obtain a specific knowledge sub-graph; and bidirectional association analysis is performed through the specific knowledge sub-graph.

[0009] As a preferred solution of the chronic disease collaborative management method of knowledge and data fusion of the present invention, wherein: the combination of the bidirectional relationship derivation method to calculate the association strength between the supplementary node and the basic node includes: calculating the association strength between the supplementary node and the basic node by the following formula:

[0010]

[0011] in, Indicates supplementary node and base nodes The strength of association, To enhance the feature dataset Item eigenvalue, is the knowledge weight associated with the corresponding feature and node, is the total number of features; when the association strength is greater than the set threshold, an associated edge is generated, and a supplementary node is added to the comprehensive knowledge graph.

[0012] As a preferred solution of the chronic disease collaborative management method of knowledge and data fusion described in the present invention, wherein: through the specific knowledge sub-graph, a bidirectional association analysis is performed, including: the bidirectional association analysis includes forward deduction and reverse verification; the forward deduction includes: according to the weights of specific nodes and edges, the patient's potential health risks are deduced to generate a specific health risk assessment value; the reverse verification includes: based on the forward deduction results, the rationality of low-weight nodes and edges in the graph is verified, and the structure of the specific knowledge sub-graph is dynamically adjusted.

[0013] As a preferred solution of the chronic disease collaborative management method of knowledge and data fusion of the present invention, the calculation formula of the health risk assessment value is as follows:

[0014]

[0015] in, For Node and The health risk assessment value between is the edge weight between two nodes, and Node and The weight of To connect to a node The total number of edges, is the depth of the association path; based on the forward deduction results, the rationality of the low-weight nodes and edges in the graph is verified, including: dividing the weights of the specific knowledge sub-graph into intervals; dividing the risk level into low risk, medium risk and high risk according to the different intervals of the health risk assessment value; comparing the characteristic vectors of the low-risk value nodes with the characteristic vectors of the high-risk value nodes to see if there is a difference; checking whether the low-risk value nodes are consistent with the actual characteristic data of the patients; checking whether the strength of the low-weight edges is within a reasonable range, and analyzing whether there is a necessary association; if the edge weight is too low and lacks actual basis, the edge is removed; checking whether the directionality and relationship of the edge match the literature and guidelines.

[0016] As a preferred solution of the knowledge and data fusion chronic disease collaborative management method of the present invention, wherein: the use of the specific knowledge sub-graph to perform multi-time series prediction and the short-term and long-term trend prediction of chronic disease health indicators includes: combining the node weights and edge association strengths in the specific knowledge sub-graph, using the enhanced feature data set Extract characteristic sequences related to health indicators and construct time series data matrix ;based on Time series data matrix, mining the trend characteristics and periodicity of health indicators, and generating short-term prediction matrices and the long-term forecast matrix ; Among them, the short-term prediction matrix is ​​extracted through high-frequency components; the long-term prediction matrix is ​​extracted through low-frequency components.

[0017] As a preferred solution of the chronic disease collaborative management method of knowledge and data fusion described in the present invention, wherein: the combination of prediction results, the patient's health trajectory model is constructed through the state transition probability, reflecting the dynamic evolution of different health states and future risk nodes, including: according to the short-term prediction matrix and the long-term forecast matrix Calculate the health state transition probability and construct the health trajectory model as follows:

[0018]

[0019] in, Represents a collection of health states; is the state transition probability matrix:

[0020]

[0021] in, Indicates status To status The transition probability, Represents a state collection Any state in Status and status The characteristic difference is defined as:

[0022]

[0023] in, and are the characteristic values ​​of the corresponding states in short-term and long-term predictions, It is the total number of feature nodes involved in the calculation of state feature differences in the specific knowledge sub-graph; the health trajectory model is dynamically labeled for risk, the nodes in the health trajectory model are divided into three states: normal, warning, and abnormal, and the locations of potential risk nodes in the future are marked.

[0024] In a second aspect, the present invention provides a knowledge and data fusion chronic disease collaborative management system, which includes: a multi-dimensional feature acquisition module, which is used to perform multi-dimensional hierarchical acquisition of patient feature data to generate a feature data set, and use a feature gain modeling method to process different dimensional data in the feature data set to generate an enhanced feature data set;

[0025] The knowledge graph generation module is used to dynamically generate a comprehensive knowledge graph of chronic diseases by combining medical literature, clinical guidelines and patient group big data. At the same time, the comprehensive knowledge graph is supplemented with personalized nodes and weights are optimized to generate patient-specific knowledge sub-graphs and perform bidirectional correlation analysis.

[0026] The time series prediction module uses the specific knowledge sub-graph to perform multi-time series predictions and short-term and long-term trend predictions on chronic disease health indicators; and combines the prediction results to build a patient health trajectory model through state transition probabilities to reflect the dynamic evolution of different health states and future risk nodes;

[0027] The dynamic risk assessment and personalized management module is used to comprehensively assess the patient's health status based on the dynamic risk labeling results of the health trajectory model, combined with the node weights and associated edges in the specific knowledge subgraph, and generate personalized management suggestions based on the assessment results.

[0028] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the method for collaborative management of chronic diseases by integrating knowledge and data as described in the first aspect of the present invention are implemented.

[0029] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: the computer program instructions, when executed by a processor, implement the steps of the method for collaborative management of chronic diseases by integrating knowledge and data as described in the first aspect of the present invention.

[0030] The beneficial effects of the present invention are as follows: the present invention significantly improves the efficiency of chronic disease management through automated data collection and preprocessing, reduces the workload of medical staff in data management, and enables them to focus more on the clinical treatment and health management of patients. Secondly, the comprehensive knowledge graph provides a standardized reference for health status assessment, and the specific knowledge sub-graph integrates the personalized characteristics of patients into the graph through personalized supplementation and optimization, thereby enhancing the pertinence and real-time nature of the knowledge graph.

[0031] In addition, the health trajectory model of the present invention reveals the evolution law of health status through state transition probability, and dynamic risk annotation provides visualization results of health risks. It combines the knowledge graph weight to generate personalized management suggestions, provide patients with targeted intervention plans, optimize lifestyles and monitoring plans, and can issue early warnings when abnormalities are detected, which helps to detect the deterioration trend of chronic diseases in advance, so as to take timely intervention measures. This function is crucial to reducing the risk of acute attacks and improving the quality of life of patients.

[0032] Through intelligent analysis, the present invention helps to rationally allocate medical resources, give priority to patients with higher risks, improve the efficiency of medical resource utilization, and reduce overall medical costs. In addition, the collaborative management platform provided can promote cooperation between different medical institutions and medical staff, realize information sharing and resource integration, and improve the overall level of medical services. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0034] Figure 1 Flowchart of the collaborative management approach for chronic diseases based on knowledge and data fusion.

[0035] Figure 2 It is a schematic diagram of the structure of a computer device suitable for implementing the embodiments of the present application.

[0036] Figure 3 This is a structural diagram of the collaborative management system for chronic diseases that integrates knowledge and data. DETAILED DESCRIPTION

[0037] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0038] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0039] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0040] Example 1

[0041] Reference Figure 1~Figure 2 , which is the first embodiment of the present invention, and provides a method for collaborative management of chronic diseases by integrating knowledge and data, including:

[0042] S1: Perform multi-dimensional hierarchical collection of patient characteristic data to generate a characteristic data set, and use the characteristic gain modeling method to process the different dimensional data in the characteristic data set to generate an enhanced characteristic data set .

[0043] Specifically, the data collected will include but are not limited to the patient's physiological signs data, living habits data, behavior monitoring data, and real-time biological signal data. Based on the data source and collection frequency, the data will be divided into static feature data and dynamic feature data.

[0044] Furthermore, the quality of the feature data set is evaluated, missing values ​​and outliers are removed, and data gaps are filled based on sampling consistency rules.

[0045] Specifically, the normalization method is used to scale the data to a unified dimension. The specific formula is as follows:

[0046]

[0047] in, is the original data point, is the normalized data point, and are the minimum and maximum values ​​of the data respectively.

[0048] Preferably, based on the individual characteristics of patients and disease characteristics, the data is decomposed by dimensions to generate subsets at three characteristic levels: physiological, behavioral, and environmental. The principal component analysis (PCA) dimensionality reduction method is then used to extract key features to obtain a reduced-dimensional feature data set.

[0049] Specifically, first, set the feature weights and feature hierarchy priority factors, and perform principal component analysis after weighted adjustment of the data subsets to increase the sensitivity to key features:

[0050]

[0051] in, is the original data matrix after stratification, containing subsets of physiological, behavioral, and environmental features, is the feature level priority factor, which assigns priority according to the importance of the level. is the feature matrix after dimensionality reduction, is the feature weight matrix, and the specific calculation formula is as follows:

[0052]

[0053] in, Features In Dimension The weight under Features variance, is the total number of features.

[0054] It should be noted that through multi-level decomposition and reorganization, the correlation between different feature levels is strengthened, providing more efficient data input for gain modeling.

[0055] Furthermore, the features in the feature data set after dimensionality reduction are modeled to generate an enhanced feature data set:

[0056] Preferably, nonlinear mapping and feature interaction modeling are used to generate enhanced feature data sets through tensor quantization processing and multi-level nested models:

[0057]

[0058] in, After dimensionality reduction, feature data, For the global nonlinear activation function, ReLU or Softplus can be used. is the number of features retained after dimensionality reduction, It is a feature nonlinear mapping function used to improve the expressiveness of a single feature. For example, use:

[0059]

[0060] in, is the characteristic amplification factor, is the characteristic shift coefficient; is a high-order interaction function between features, constructing a nonlinear coupling relationship between feature pairs:

[0061]

[0062] The interaction results are mapped to the interval (-1, 1) through the tangent function to reflect the nonlinear correlation between features.

[0063] S2: Dynamically generate a comprehensive knowledge graph of chronic diseases by combining medical literature, clinical guidelines and patient population big data; at the same time, perform personalized node supplementation and weight optimization on the comprehensive knowledge graph to generate a patient-specific knowledge sub-graph.

[0064] Through bidirectional correlation analysis, on the one hand, the potential health risks of patient characteristics are deduced, and on the other hand, the rationality of the graph relationship is reversely verified to ensure the scientific nature of the dynamic adjustment of the graph.

[0065] It should be noted that based on medical literature, clinical guidelines and patient group big data, structured knowledge related to chronic disease management is automatically extracted to construct a knowledge graph.

[0066] S2.1: Knowledge Graph G It is a directed graph in which nodes represent medical concepts and edges represent the relationships between concepts.

[0067] It can be expressed as G =( V , E ),in V is a collection of nodes, E is the edge set, .

[0068] S2.2: Based on enhanced feature dataset Dynamically mine individual-specific features, generate supplementary nodes and associated edges, combine the bidirectional relationship derivation method, calculate the association strength between the supplementary nodes and the basic nodes, and add them to the comprehensive knowledge graph to generate a preliminary patient-specific knowledge subgraph.

[0069] Specifically, in the enhanced feature dataset Individual-specific features are extracted from the dataset, and features with significance higher than a set threshold are regarded as key features for generating supplementary nodes. The key features that meet the above conditions are grouped through a clustering algorithm (such as improved K-Means or density-based DBSCAN), and each group represents a supplementary node.

[0070] Preferably, the association strength between the supplementary node and the basic node is calculated by the following formula:

[0071]

[0072] in, Indicates supplementary node and base nodes The strength of association, To enhance the feature dataset Item eigenvalue, is the knowledge weight associated with the corresponding feature and node, is the total number of features.

[0073] When the association strength is greater than the set threshold, an associated edge is generated and the supplementary node is added to the comprehensive knowledge graph.

[0074] S2.3: According to the characteristics of nodes and edges, a weight adjustment method based on feature contribution rate is used to optimize the preliminary specific knowledge subgraph to obtain a specific knowledge subgraph.

[0075] Specifically, let each node The knowledge weight is , by patient data feature importance and the strength of association Dynamically adjust weights:

[0076]

[0077] in, and Assign parameters to the weights, For Node The set of neighbor nodes.

[0078] S2.4: Perform bidirectional association analysis through specific knowledge sub-graphs:

[0079] Forward deduction: Based on the weights of specific nodes and edges, the patient's potential health risks are deduced and specific health risk assessment values ​​are generated;

[0080] Reverse verification: Based on the forward derivation results, the rationality of the low-weight nodes and edges in the graph is verified, and the structure of the specific knowledge sub-graph is dynamically adjusted to ensure the scientific nature of the patient-specific knowledge sub-graph.

[0081] Among them, the potential health risk of the patient is derived, and the health risk assessment value is calculated as follows:

[0082]

[0083] in, For Node and The health risk assessment value between is the edge weight between two nodes, and Node and The weight of To connect to a node The total number of edges, is the depth of the associated path, that is, the maximum number of jumps allowed to expand when inferring along the edge starting from the current node in the specific knowledge subgraph.

[0084] According to the calculated Values ​​are used to classify risk levels into low risk, medium risk, and high risk:

[0085] Divide the graph weight into intervals;

[0086] like The value is in the lowest quantile interval of the atlas weight, indicating that the patient's current health condition is relatively stable and is recorded as low risk;

[0087] like Values ​​in the median interval of the atlas weight indicate that there are certain problems with the health status and need to be closely monitored, which is recorded as medium risk;

[0088] like Values ​​in the highest percentile interval of the spectrum weight indicate abnormal health conditions and require immediate intervention, recorded as high risk.

[0089] Further, reverse verification includes: The nodes and edges corresponding to the low risk values ​​are rationally verified, and the structure of the specific knowledge sub-graph is dynamically adjusted, including:

[0090] Compare whether there are significant differences between the feature vectors of low-risk value nodes and high-risk value nodes; check whether the low-risk value nodes are consistent with the actual feature data of the patient to avoid deviation between the atlas node features and the actual situation; if the node does not match the actual data, re-label or update the features.

[0091] Check whether the strength of low-weight edges is within a reasonable range and analyze whether there is a necessary association; if the edge weight is too low and lacks practical basis, remove the edge; check whether the directionality and relationship of the edge match the literature and guidelines.

[0092] Ultimately, unreasonable nodes and edges are removed or weakened to avoid affecting the overall scientific nature of the graph; the nodes and edges of the graph are re-optimized to ensure that the high-weight parts are more significant and support health risk assessment, and the adjusted graph is re-verified to ensure that the optimized graph can better reflect the patient's health risk characteristics.

[0093] It can be seen that by introducing the bidirectional association analysis method, the knowledge graph can not only support the reasoning of patients' health status, but also optimize the dynamic adjustment of the graph through patient data.

[0094] S3: Use specific knowledge sub-graphs to perform multi-time series predictions and make short-term and long-term trend predictions for chronic disease health indicators; and combine the prediction results to build a patient health trajectory model through state transition probability to intuitively reflect the dynamic evolution of different health states and future risk nodes.

[0095] S3.1: Combining the node weights and edge association strengths in the specific knowledge subgraph, using the enhanced feature dataset Extract characteristic sequences related to health indicators and construct time series data matrix .

[0096] in, It is a matrix with time steps as rows and eigenvalues ​​as columns, which is used to dynamically describe the time series change characteristics of patient health indicators.

[0097] Specifically, nodes whose node weights are higher than a set threshold are selected as candidate nodes for health indicator-related features; among the candidate nodes, nodes whose associated edge strengths are higher than a set threshold are further selected to ultimately form a core feature node set.

[0098] Map the time steps in the enhanced feature dataset to the features in the feature node set one by one;

[0099] For each time step, extract the eigenvalues ​​of the corresponding core feature node set in the enhanced feature data set to form an eigenvalue vector;

[0100] The eigenvalue vectors extracted at each time step are stacked sequentially as matrix rows to form a time series data matrix.

[0101] S3.2: Based on Time series data matrix, mining the trend characteristics and periodicity of health indicators, and generating short-term prediction matrices and the long-term forecast matrix .

[0102] Among them, the short-term prediction matrix is ​​extracted through high-frequency components to indicate the changing trend of health indicators in the short term in the future; the long-term prediction matrix is ​​extracted through low-frequency components to characterize the long-term fluctuations and potential risk points of health indicators.

[0103] S3.3: Based on the short-term and long-term prediction matrices combined with the state transition probability model, the transition probabilities between different health states are calculated to generate the patient health trajectory model, where the patient's health trajectory model uses a directed graph to represent the health state and its transition relationship, where the node is the health state and the edge weight is the state transition probability.

[0104] Better, according to the prediction matrix and , calculate the health state transition probability and construct the health trajectory model as follows:

[0105]

[0106] in, Represents a set of health states; is the state transition probability matrix:

[0107]

[0108] in, Indicates status To status The transition probability, Represents a state collection Any state in Status and status The characteristic difference is defined as:

[0109]

[0110] in, and are the characteristic values ​​of the corresponding states in short-term and long-term predictions, The total number of feature nodes involved in the calculation of state feature differences in the specific knowledge subgraph.

[0111] S3.4: Dynamically label the health trajectory model for risks, classify the nodes in the health trajectory model into three states: normal, warning, and abnormal, and mark the locations of future potential risk nodes to intuitively reflect the dynamic evolution of the patient's health status.

[0112] S3.5: Interactively associate the patient's health trajectory model with the specific knowledge sub-graph, dynamically optimize the weight and path structure of state transitions in the trajectory model, and ensure that the health trajectory model can accurately reflect the patient's personalized health status and its evolution trend.

[0113] S4: Based on the dynamic risk labeling results of the health trajectory model, combined with the node weights and associated edges in the specific knowledge subgraph, the patient's health status is comprehensively evaluated, and personalized management recommendations are generated based on the evaluation results.

[0114] Among them, personalized management recommendations include intervention measures, lifestyle optimization and regular monitoring programs to achieve comprehensive management and continuous improvement of patients' health status.

[0115] Specifically, based on the dynamic labeling results of each node in the health trajectory model, the health status is divided into three categories: normal, warning and abnormal:

[0116] For nodes marked as normal, identify key characteristic nodes and their associated edges in the state, confirm health stability and the effectiveness of current management strategies;

[0117] For nodes marked as alert, the high-frequency change characteristics in the short-term prediction matrix are combined to evaluate the causes of health indicator deviations and potential intervention windows;

[0118] For nodes marked as abnormal, the potential risk features in the long-term prediction matrix are used to analyze the impact of key nodes and edges, and mark priority processing areas.

[0119] The patient's health status is evaluated through the state transition probability matrix of the health trajectory model combined with the node weights and edge strengths in the specific knowledge subgraph.

[0120] Among them, the contribution rate of the key characteristic nodes of each health state to the risk assessment is calculated, and a comprehensive evaluation is performed according to the following rules:

[0121] Changes in high-weight nodes and their associated edges have a higher priority; edges with high state transition probabilities mark the main paths for changes in patients' health status; combining short-term and long-term trend predictions of health indicators, possible high-risk transfer nodes and paths in the future are identified.

[0122] Furthermore, based on the health status assessment results, we provide patients with personalized management suggestions in the following three aspects:

[0123] For nodes in abnormal states, enhanced intervention measures targeting key characteristic nodes are recommended, such as medication adjustment, medical intervention, or other targeted treatments.

[0124] For nodes in the alert state, mild intervention strategies are proposed, such as increasing exercise, adjusting diet, or psychological counseling.

[0125] Based on the node associations in the specific knowledge subgraph, optimization suggestions are made for daily behavior patterns, such as reducing the intake of unhealthy foods, increasing exercise frequency, or adjusting work and rest schedules.

[0126] Patients are advised to improve high-risk factors in environmental characteristics, such as reducing pollution exposure or improving living environment.

[0127] And set a monitoring frequency for each health status. For example, the alert status requires a monthly health review, and the "abnormal" status requires weekly or daily monitoring of core health indicators.

[0128] This embodiment also provides a computer device, which is suitable for the method of collaborative management of chronic diseases by integrating knowledge and data, and includes a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the method of collaborative management of chronic diseases by integrating knowledge and data as proposed in the above embodiment.

[0129] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0130] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for collaborative management of chronic diseases by realizing knowledge and data fusion as proposed in the above embodiment is implemented.

[0131] Example 2

[0132] Reference Figure 3 , which is the second embodiment of the present invention, and provides a knowledge and data fusion chronic disease collaborative management system, including:

[0133] A multi-dimensional feature acquisition module, used for performing multi-dimensional hierarchical acquisition of patient feature data, generating a feature data set, and processing different dimensional data in the feature data set using a feature gain modeling method to generate an enhanced feature data set;

[0134] The knowledge graph generation module is used to dynamically generate a comprehensive knowledge graph of chronic diseases by combining medical literature, clinical guidelines and patient group big data. At the same time, the comprehensive knowledge graph is supplemented with personalized nodes and weights are optimized to generate patient-specific knowledge sub-graphs and perform bidirectional correlation analysis.

[0135] The time series prediction module uses the specific knowledge sub-graph to perform multi-time series predictions and short-term and long-term trend predictions on chronic disease health indicators; and combines the prediction results to build a patient health trajectory model through state transition probabilities to reflect the dynamic evolution of different health states and future risk nodes;

[0136] The dynamic risk assessment and personalized management module is used to comprehensively assess the patient's health status based on the dynamic risk labeling results of the health trajectory model, combined with the node weights and associated edges in the specific knowledge subgraph, and generate personalized management suggestions based on the assessment results.

[0137] In summary, the present invention significantly improves the efficiency of chronic disease management through automated data collection and preprocessing, reduces the workload of medical staff in data management, and enables them to focus more on the clinical treatment and health management of patients. Secondly, the comprehensive knowledge graph provides a standardized reference for health status assessment, and the specific knowledge sub-graph integrates the patient's personalized characteristics into the graph through personalized supplementation and optimization, thereby enhancing the pertinence and real-time nature of the knowledge graph.

[0138] In addition, the health trajectory model of the present invention reveals the evolution law of health status through state transition probability, and dynamic risk annotation provides visualization results of health risks. It combines the knowledge graph weight to generate personalized management suggestions, provide patients with targeted intervention plans, optimize lifestyles and monitoring plans, and can issue early warnings when abnormalities are detected, which helps to detect the deterioration trend of chronic diseases in advance, so as to take timely intervention measures. This function is crucial to reducing the risk of acute attacks and improving the quality of life of patients.

[0139] Through intelligent analysis, the present invention helps to rationally allocate medical resources, give priority to patients with higher risks, improve the efficiency of medical resource utilization, and reduce overall medical costs. In addition, the collaborative management platform provided can promote cooperation between different medical institutions and medical staff, realize information sharing and resource integration, and improve the overall level of medical services.

[0140] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for collaborative management of chronic diseases by integrating knowledge and data, characterized by: include: Perform multi-dimensional hierarchical collection of patient characteristic data to generate a characteristic data set, and use a characteristic gain modeling method to process data of different dimensions in the characteristic data set to generate an enhanced characteristic data set; Combine medical literature, clinical guidelines and patient group big data to dynamically generate a comprehensive knowledge graph of chronic diseases. At the same time, perform personalized node supplementation and weight optimization on the comprehensive knowledge graph to generate patient-specific knowledge sub-graphs and conduct bidirectional correlation analysis. The specific knowledge sub-graph is used to perform multi-time series prediction, and short-term and long-term trend predictions of chronic disease health indicators are performed; and combined with the prediction results, a patient health trajectory model is constructed through state transition probability to reflect the dynamic evolution of different health states and future risk nodes; According to the dynamic risk labeling results of the health trajectory model, combined with the node weights and associated edges in the specific knowledge subgraph, the patient's health status is comprehensively evaluated, and personalized management suggestions are generated based on the evaluation results; The bidirectional association analysis performed through the specific knowledge sub-graph includes: the bidirectional association analysis includes forward deduction and reverse verification; the forward deduction includes: according to the weights of specific nodes and edges, deriving the patient's potential health risks and generating specific health risk assessment values; the reverse verification includes: based on the forward deduction results, verifying the rationality of low-weight nodes and edges in the graph, and dynamically adjusting the structure of the specific knowledge sub-graph.

2. The method for collaborative management of chronic diseases by integrating knowledge and data according to claim 1, characterized in that: The following steps are involved in supplementing the comprehensive knowledge graph with personalized nodes and optimizing its weights, generating a patient-specific knowledge sub-graph, and performing a bidirectional association analysis: Based on enhanced feature dataset Dynamically mining individual-specific features, generating supplementary nodes and associated edges, combining a bidirectional relationship derivation method, calculating the association strength between the supplementary nodes and the basic nodes, and adding them to the comprehensive knowledge graph to generate a preliminary patient-specific knowledge subgraph; According to the characteristics of nodes and edges, a weight adjustment method based on feature contribution rate is adopted to optimize the preliminary specific knowledge sub-graph to obtain the specific knowledge sub-graph; Bidirectional association analysis is performed through the specific knowledge sub-graph.

3. The method for collaborative management of chronic diseases by integrating knowledge and data according to claim 2, characterized in that: The method for inferring a bidirectional relationship and calculating the association strength between the supplementary node and the basic node includes: The strength of association between the supplementary node and the base node is calculated by the following formula: ， in, Indicates supplementary node and base nodes The strength of association, To enhance the feature dataset Item eigenvalue, is the knowledge weight associated with the corresponding feature and node, is the total number of features; When the association strength is greater than a set threshold, an association edge is generated and a supplementary node is added to the comprehensive knowledge graph.

4. The method for collaborative management of chronic diseases by integrating knowledge and data according to claim 3, characterized in that: The calculation formula of the health risk assessment value is as follows: ， in, For Node and The health risk assessment value between is the edge weight between two nodes, and Node and The weight of To connect to a node The total number of edges, is the depth of the association path; Based on the forward derivation results, the rationality verification of low-weight nodes and edges in the graph includes: Divide the weights of specific knowledge sub-graphs into intervals; According to the different intervals of health risk assessment values, the risk levels are divided into low risk, medium risk and high risk; Compare whether there is a difference between the feature vector of the low risk value node and the feature vector of the high risk value node; Check whether the low-risk value node is consistent with the patient's actual characteristic data; Check whether the strength of low-weight edges is within a reasonable range and analyze whether there is a necessary association; if the edge weight is too low and lacks practical basis, remove the edge; check whether the directionality and relationship of the edge match the literature and guidelines.

5. The method for collaborative management of chronic diseases by integrating knowledge and data according to claim 1, characterized in that: The use of the specific knowledge sub-graph to perform multi-time series prediction and short-term and long-term trend prediction of chronic disease health indicators includes: Combining the node weights and edge association strengths in the specific knowledge subgraph, using the enhanced feature dataset Extract characteristic sequences related to health indicators and construct time series data matrix ; based on Time series data matrix, mining the trend characteristics and periodicity of health indicators, and generating short-term prediction matrices and the long-term forecast matrix ; Among them, the short-term prediction matrix is ​​extracted through high-frequency components; the long-term prediction matrix is ​​extracted through low-frequency components.

6. The method for collaborative management of chronic diseases by integrating knowledge and data according to claim 1, characterized in that: The above combined prediction results and the state transition probability are used to construct the patient's health trajectory model, which reflects the dynamic evolution of different health states and future risk nodes, including: According to the short-term forecast matrix and the long-term forecast matrix Calculate the health state transition probability and construct the health trajectory model as follows: ， in, Represents a collection of health states; is the state transition probability matrix: ， in, Indicates status To status The transition probability, Represents a state collection Any state in Status and status The characteristic difference is defined as: ， in, and are the characteristic values ​​of the corresponding states in short-term and long-term predictions, The total number of feature nodes participating in the state feature difference calculation in the specific knowledge subgraph; Dynamically label the health trajectory model for risks, classify the nodes in the health trajectory model into three states: normal, warning, and abnormal, and mark the locations of future potential risk nodes.

7. A knowledge and data fusion chronic disease collaborative management system, based on the knowledge and data fusion chronic disease collaborative management method according to any one of claims 1 to 6, characterized in that: Also includes: A multi-dimensional feature acquisition module, used for performing multi-dimensional hierarchical acquisition of patient feature data, generating a feature data set, and processing different dimensional data in the feature data set using a feature gain modeling method to generate an enhanced feature data set; The knowledge graph generation module is used to dynamically generate a comprehensive knowledge graph of chronic diseases by combining medical literature, clinical guidelines and patient group big data. At the same time, the comprehensive knowledge graph is supplemented with personalized nodes and weights are optimized to generate patient-specific knowledge sub-graphs and perform bidirectional correlation analysis. The time series prediction module uses the specific knowledge sub-graph to perform multi-time series predictions and short-term and long-term trend predictions on chronic disease health indicators; and combines the prediction results to build a patient health trajectory model through state transition probabilities to reflect the dynamic evolution of different health states and future risk nodes; The dynamic risk assessment and personalized management module is used to comprehensively assess the patient's health status based on the dynamic risk labeling results of the health trajectory model, combined with the node weights and associated edges in the specific knowledge subgraph, and generate personalized management suggestions based on the assessment results.

Images