Dynamic risk network modeling and real-time comprehensive risk assessment method and system for elderly multi-disease co-prevalent population

Abstract

The application provides a dynamic risk network modeling and real-time comprehensive risk assessment method and system for the elderly multi-disease co-prevalent population, relates to the data modeling and monitoring technical field of multiple diseases, and comprises the following steps: constructing a population-level baseline risk network based on historical medical data, which is used to reflect the average disease condition of multiple patients; constructing an individual risk network for an individual patient, which is used to reflect the disease condition of the individual patient; periodically updating the individual risk network dynamically, and performing dynamic risk assessment based on the population-level baseline risk network and the individual risk network constructed for the individual patient; and generating individualized intervention suggestions according to the dynamic risk assessment results. The application has the advantages that the correlation between multiple diseases and the comprehensive risk can be expressed in a structured manner, and the evaluation results with strong interpretability can be output, so that the continuous health monitoring level of the elderly multi-disease co-prevalent population is improved.

Description

Technical Field

[0001] This invention relates to the field of data modeling and monitoring technology for various diseases, and more specifically, to a method and system for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities. Background Technology

[0002] Elderly individuals commonly suffer from multiple comorbidities (such as hypertension, diabetes, coronary heart disease, chronic kidney disease, and COPD) and exhibit long-term fluctuations in various indicators. Routine monitoring often requires considering the risk of exacerbation due to the interaction of these multiple diseases.

[0003] Current technologies for monitoring vital signs in elderly individuals with multiple comorbidities typically utilize multiple single-disease risk assessment models, each with its own predictive model for each specific disease. This approach results in fragmented calculations, making it difficult to generate unified risk quantification indicators and effectively express the interactions between different diseases and between diseases and indicators. However, in comorbid scenarios, synergistic or additive effects often exist between diseases; for example, the worsening of one disease may significantly increase the risk of another. Current technologies struggle to effectively monitor such situations, and the interpretability of the output results is weak.

[0004] Therefore, we should continue to improve the monitoring methods for people with multiple diseases, express the associations and comprehensive risks among multiple diseases in a structured manner, and output highly interpretable assessment results, thereby improving the level of continuous health monitoring for elderly people with multiple diseases. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities. It realizes a unified expression of the correlation and comprehensive risk among multiple diseases in a structured manner and outputs assessment results with strong interpretability.

[0006] This invention is achieved through the following technical solution: A method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly individuals with multiple comorbidities includes the following steps: A population-level baseline risk network is constructed based on historical medical data to reflect the average disease status of multiple patients. It includes disease nodes, disease parameters, directed edges, and baseline edge weights. Each disease node represents a disease type. Disease parameters that characterize disease-related vital signs are set on the disease nodes. The influence relationship between disease nodes is represented by directed edges. The baseline edge weights represent the strength of the influence relationship between upstream disease nodes and downstream disease nodes. An individual risk network is constructed for each individual patient to reflect their disease status. This network includes individual disease nodes, individual disease parameters, hidden nodes, directed edges, and risk edge weights. Each individual disease node represents a disease type possessed by the individual patient. Individual disease parameters are set on each individual disease node to represent disease-related vital signs. The directed edge connections between individual diseases are consistent with the population-level baseline risk network. Risk edge weights are used to indicate the probability that real-time data from upstream individual disease nodes will induce or cause the deterioration of disease parameters in downstream disease nodes. Hidden nodes are nodes that are not individual disease nodes but belong to disease nodes and are downstream nodes of the disease nodes corresponding to the individual disease nodes. The individual risk network is periodically updated dynamically, and an individual risk network is constructed based on the population-level baseline risk network and individual patients for dynamic risk assessment. Individualized intervention recommendations are generated based on the results of dynamic risk assessment.

[0007] Preferably, the method for constructing the population-level baseline risk network is as follows: Access historical medical data and medical knowledge databases for multiple patients; Identify disease milestones based on historical medical data; Based on the average value of disease-related vital signs data of all patients in historical medical data, disease parameters are assigned to each disease node. The influence relationships between disease nodes are determined based on a medical knowledge database, and corresponding directed edges are established. The edge weights of each directed edge are calculated based on the historical medical data.

[0008] Preferably, assuming the existence of the directed edge from disease node i to disease node j, the method for calculating the baseline edge weight based on the historical medical data is as follows: Calculate the probability parameter of coexistence of the diseases corresponding to disease node i and disease node j in the same patient based on historical medical data. and ratio ratio parameter : ;

[0009] ; in, This represents the percentage of all patients with disease node i in the historical medical data who also have disease node j. This represents the number of patients in the historical medical data who have both disease node i and disease node j. This represents the number of patients in the historical medical data who have disease node i but do not have disease node j. This represents the number of patients in the historical medical data who do not have disease node i but do have disease node j. This represents the number of patients in the historical medical data who do not have disease node i and disease node j. The ratio between disease node i and disease node j. and These are the first threshold and the second threshold, respectively, with the second threshold being greater than the first threshold; Calculate the baseline edge weights of the directed edges from disease node i to disease node j in the population-level baseline risk network. : .

[0010] Preferably, the method for constructing the individual risk network is as follows: Obtain the population-level baseline risk network, retain the disease nodes corresponding to the diseases of individual patients and the downstream disease nodes of the disease nodes corresponding to the existing diseases, and delete other nodes; Traverse the remaining disease nodes. If an individual patient does not have the disease, define the disease node as a hidden node; otherwise, define it as an individual disease node. Based on the diagnostic data, the disease parameters of the individual disease node are modified to the disease-related vital signs data of the individual patient; Assume there exists a directed edge from individual disease node i to individual disease node j, and the risk edge weight is... The calculation method is as follows: ; ; in, Let be the baseline edge weight of the directed edge from disease node i to disease node j in the population-level baseline risk network, where disease node i' in the population-level baseline risk network and individual disease node i' in the individual risk network are of the same disease, and the value of i' is within the range of all disease nodes in the population-level baseline risk network. `clip` is used to... A function that clips values ​​to [0,1]. For intermediate parameters, It is an adjustment coefficient and is a positive number. This represents the deviation of the k-th disease parameter of individual disease node i from its corresponding normal value. It is the average value of the deviation of the k-th disease parameter of disease node i in the population-level baseline risk network from the corresponding normal value.

[0011] Preferably, the method for dynamically updating the individual risk network is as follows: Obtain the most recent diagnostic data, update the disease parameters and the disease node, and re-apply according to the risk edge weights. Calculation method .

[0012] Preferably, the method for dynamic risk assessment based on the population-level baseline risk network and the individual risk network is as follows: The disease-inducing risk of an individual patient is obtained based on the risk edge weight and the baseline edge weight; Individual patient health risk is obtained based on the disease parameters of the individual risk network.

[0013] Preferably, the method for obtaining the disease-inducing risk of the individual patient based on the risk edge weights and the baseline edge weights is as follows: Obtain the individual disease node i in the individual risk network, and obtain all its upstream nodes in the individual risk network. Calculate the disease association worsening risk assessment parameter for individual disease node i. : ; Among them, individual disease node b is an individual disease node. The upstream node, and These are individual disease node b to individual disease node b. The risk edge weights and their corresponding baseline edge weights in the population-level baseline risk network. For individual disease nodes The total number of upstream nodes; Obtain hidden nodes in the individual risk network And obtain all upstream nodes in the individual risk network, and calculate hidden nodes. Disease-associated risk assessment parameters : ; Among them, hidden nodes Corresponding individual disease nodes Individual disease nodes For individual disease nodes The upstream node, and Individual disease nodes To individual disease nodes The risk edge weights and their corresponding baseline edge weights in the population-level baseline risk network. For individual disease nodes The total number of upstream nodes.

[0014] Preferably, the health risk of an individual patient is obtained based on the disease parameters of the individual risk network. The method is as follows: ; ; in, This represents the health risk of the disease corresponding to individual disease node i. This represents the deviation of the k-th disease parameter of individual disease node i from its corresponding normal value. The total number of disease parameters representing individual disease node i. This represents the total number of individual disease nodes.

[0015] Preferably, the method for generating individualized intervention recommendations based on dynamic risk assessment results is as follows: If the disease-related worsening risk assessment parameters If the value is greater than the corresponding preset threshold, an individual disease node is issued. Warning of the risk of complications and worsening of the corresponding disease; If the disease-associated risk assessment parameters If the value is greater than the corresponding preset threshold, then a hidden node is emitted. Warning of the risk of complications from corresponding diseases; If individual patient's health risk If the value is greater than the corresponding preset threshold, an individual patient's health indicator risk warning will be issued.

[0016] This invention also summarizes a dynamic risk network modeling and real-time comprehensive risk assessment system for elderly people with multiple comorbidities, applied to the aforementioned dynamic risk network modeling and real-time comprehensive risk assessment method for elderly people with multiple comorbidities, including: The population-level baseline risk network construction module is used to build a population-level baseline risk network based on historical medical data, which reflects the average disease status of multiple patients. The Individual Risk Network Construction Module is used to build an individual risk network for each individual patient to reflect their disease status. The dynamic update module is used to periodically update the individual risk network. The risk assessment module is used to conduct dynamic risk assessment by constructing individual risk networks based on population-level baseline risk networks and individual patients. The results generation module is used to generate individualized intervention recommendations based on the results of dynamic risk assessment.

[0017] The technical solution of the present invention has at least the following advantages and beneficial effects: This invention establishes two networks: a population-level baseline risk network and an individual risk network. The population-level baseline risk network provides basic information on diseases and their interrelationships, while the individual risk network reflects the real-time status of patients through dynamic updates. By combining and comparing the two networks, risk assessment is performed, thereby improving the interpretability of risk assessment. This invention integrates multiple diseases and their interactions into a single model, enabling a unified quantitative assessment of the overall risk and disease interactions of patients with multiple comorbidities through a single model. This invention expresses the degree of influence between diseases through directed edges and edge weights, and adopts a dynamic edge weight update mechanism in the individual risk network. This enables the depiction of the deterioration or complication risk caused by the current interaction between diseases in patients, overcoming the problem that traditional linear models are difficult to express complex interaction relationships, improving the refinement of disease risk assessment, and enhancing the interpretability of interaction relationship monitoring. This invention is rationally designed and enables unified modeling, real-time assessment, and interpretable analysis of the risk of multiple comorbidities, thereby significantly improving the precision of patient data management and clinical decision support capabilities for patients with multiple comorbidities. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the dynamic risk network modeling and real-time comprehensive risk assessment method for elderly people with multiple comorbidities provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram of the structure of the dynamic risk network modeling and real-time comprehensive risk assessment system for elderly people with multiple comorbidities provided in Embodiment 2 of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Example 1 This embodiment provides a method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly individuals with multiple comorbidities. (See also...) Figure 1 This includes the following steps: Step S1: Construct two types of risk networks: a population-level baseline risk network and an individual risk network. These two risk networks respectively represent the population statistical patterns of elderly patients and the real-time health status of individuals, and provide a structural basis for subsequent dynamic risk assessment. Specifically: On the one hand, a population-level baseline risk network is constructed based on historical medical data to reflect the average disease status of multiple patients, including disease nodes, disease parameters, directed edges, and baseline edge weights. Each disease node represents a disease type, and disease parameters representing disease-related vital signs are set on the disease nodes. The influence relationship between disease nodes is represented by directed edges, and the baseline edge weights represent the strength of the influence relationship between upstream disease nodes and downstream disease nodes.

[0021] In this embodiment, the method for constructing the population-level baseline risk network is as follows: Access historical medical data and medical knowledge databases for multiple patients; Identify disease milestones based on historical medical data; Based on the average value of disease-related vital signs data of all patients in historical medical data, disease parameters are assigned to each disease node. The influence relationships between disease nodes are determined based on a medical knowledge database, and corresponding directed edges are established. The edge weights of each directed edge are calculated based on the historical medical data.

[0022] Based on the above scheme, assuming the existence of the directed edge from disease node i to disease node j, the method for calculating the baseline edge weight based on the historical medical data is as follows: Calculate the probability parameter of coexistence of the diseases corresponding to disease node i and disease node j in the same patient based on historical medical data. and ratio ratio parameter : ;

[0023] ; in, This represents the percentage of all patients with disease node i in the historical medical data who also have disease node j. This represents the number of patients in the historical medical data who have both disease node i and disease node j. This represents the number of patients in the historical medical data who have disease node i but do not have disease node j. This represents the number of patients in the historical medical data who do not have disease node i but do have disease node j. This represents the number of patients in the historical medical data who do not have disease node i and disease node j. The ratio between disease node i and disease node j. and These are the first threshold and the second threshold, respectively, with the second threshold being greater than the first threshold; Calculate the baseline edge weights of the directed edges from disease node i to disease node j in the population-level baseline risk network. : .

[0024] In the above schemes, To calculate the number of disease nodes based on historical data Disease nodes under the corresponding disease premise Disease nodes The probability of coexistence within the same patient is used to reflect the occurrence of disease nodes. When dealing with a corresponding disease, disease nodes will also appear. The probability of the corresponding disease. In the calculation, disease nodes are represented by ratios. The corresponding disease The corresponding amplification effect of the disease was investigated, and normalization was achieved through a piecewise function to eliminate extreme values. Finally, for... and Weighted fusion is performed to obtain the baseline edge weight, which can quantify the average influence of the disease of the upstream node of a directed edge on the disease of the downstream node.

[0025] On the other hand, an individual risk network is constructed for individual patients to reflect their disease status. This network includes individual disease nodes, individual disease parameters, hidden nodes, directed edges, and risk edge weights. Each individual disease node represents a disease type possessed by the individual patient. Individual disease parameters are set on the individual disease nodes to represent disease-related vital signs. The directed edge connections between individual diseases are consistent with the population-level baseline risk network. Risk edge weights are used to indicate the probability that real-time data from upstream individual disease nodes will induce or cause the deterioration of disease parameters in downstream disease nodes. Hidden nodes are nodes that are not individual disease nodes but belong to disease nodes and are downstream nodes of the disease nodes corresponding to the individual disease nodes.

[0026] As a preferred embodiment, the method for constructing the individual risk network is as follows: The population-level baseline risk network is obtained, and the disease nodes corresponding to the diseases of individual patients and the downstream disease nodes of the disease nodes corresponding to the existing diseases are retained. Other nodes are deleted. This step is to retain the disease nodes corresponding to the diseases of individual patients and all their downstream disease nodes on the basis of the population-level network, thereby forming a sub-network structure related to the individual. After removing disease nodes irrelevant to individual patients, the remaining disease nodes need to be categorized. This is because current disease nodes fall into two categories: those representing diseases the individual patient suffers from, and those representing diseases the individual patient does not have, but which are downstream nodes of the corresponding disease in the population-level baseline risk network. Therefore, it's necessary to traverse the remaining disease nodes. If an individual patient does not have the disease, the disease node is defined as a hidden node; otherwise, it's an individual disease node. Hidden nodes are used to characterize potential risk evolution paths, while individual disease nodes are used to characterize actual disease status.

[0027] Based on the diagnostic data, the disease parameters of individual disease nodes are modified to the disease-related vital signs data of the individual patient, thereby realizing individualized modeling of disease-related data.

[0028] In constructing an individual risk network, the risk edge weights are calculated. The method introduces a dynamic adjustment mechanism based on the degree of individual anomaly on the baseline edge weights. It assumes the existence of a directed edge from individual disease node i to individual disease node j, and the risk edge weights... The calculation method is as follows: ; ; in, Let be the baseline edge weight of the directed edge from disease node i to disease node j in the population-level baseline risk network, where disease node i' in the population-level baseline risk network and individual disease node i' in the individual risk network are of the same disease, and the value of i' is within the range of all disease nodes in the population-level baseline risk network. `clip` is used to... A function that clips values ​​to [0,1]. For intermediate parameters, It is an adjustment coefficient and is a positive number. This represents the deviation of the k-th disease parameter of individual disease node i from its corresponding normal value. It is the average value of the deviation of the k-th disease parameter of disease node i in the population-level baseline risk network from the corresponding normal value.

[0029] The calculations in the above schemes The principle is to calculate the deviation of each disease parameter at an individual disease node from the normal value and compare it with the average deviation of the population to obtain the relative degree of abnormality. This allows the risk edge weights to dynamically reflect the individual patient's current position at the disease node. The impact of the corresponding disease state on the disease risk transmission capacity is used to calculate the basic edge weights. (i.e., baseline edge weights) and relative anomaly degree All of them are based on big data of the population.

[0030] Based on the processing in step S1, the population-level baseline risk network can provide a stable population statistical structure, while the individual risk network can provide a real-time representation of individual states. In particular, the hidden nodes can extend the risk network with potential risk paths. The combined representation of these two risk networks can balance the model's generalization and personalization capabilities.

[0031] Step S2: Periodically update the individual risk network dynamically, using the following method: Obtain the most recent diagnostic data, update the disease parameters and the disease node, and re-apply according to the risk edge weights. Calculation method .

[0032] Dynamic updates primarily involve constructing an individual risk network that best represents the current state of each individual patient based on the latest diagnosis.

[0033] Step S3: This step is to construct an individual risk network based on the population-level baseline risk network and the individual patient to conduct dynamic risk assessment, and to conduct a comprehensive risk assessment of the individual patient from the perspectives of disease association and individual health status.

[0034] The individual patient's disease-inducing risk is obtained based on the risk edge weights and the baseline edge weights, using the following method: Obtain the individual disease node i in the individual risk network, and obtain all its upstream nodes in the individual risk network. Calculate the disease association worsening risk assessment parameter for individual disease node i. : ; Among them, individual disease node b is an individual disease node. The upstream node, and These are individual disease node b to individual disease node b. The risk edge weights and their corresponding baseline edge weights in the population-level baseline risk network. For individual disease nodes The total number of upstream nodes.

[0035] This characterizes the upstream disease's impact on disease nodes under the current individual state. The degree to which the impact of the corresponding disease is amplified relative to the average level of the population.

[0036] Obtain hidden nodes in the individual risk network And obtain all upstream nodes in the individual risk network, and calculate hidden nodes. Disease-associated risk assessment parameters : ; Among them, hidden nodes Corresponding individual disease nodes Individual disease nodes For individual disease nodes The upstream node, and Individual disease nodes To individual disease nodes The risk edge weights and their corresponding baseline edge weights in the population-level baseline risk network. For individual disease nodes The total number of upstream nodes.

[0037] Used to detect hidden nodes in patients who do not have the disease. When dealing with a specific disease, the risk trend of the disease being induced by the corresponding upstream node is assessed by comparing the change in the risk edge weight from the upstream node to the hidden node with the baseline edge weight.

[0038] This step also quantifies the patient's current health status based on the degree of deviation between individual disease parameters and normal values, that is, obtaining the individual patient's health risk based on the disease parameters of the individual risk network. The method is as follows: ; ; in, This represents the health risk associated with the disease corresponding to individual disease node i, reflecting the degree of abnormality of the single disease at the current moment. This represents the deviation of the k-th disease parameter of individual disease node i from its corresponding normal value. The total number of disease parameters representing individual disease node i. This represents the total number of individual disease nodes. This involves summarizing all individual disease nodes to assess the overall health risk of the individual.

[0039] Through the above calculations, step S3 establishes the enhanced risk transmission between diseases driven by changes in edge weights, and the degree of abnormality in current health status driven by deviations in disease parameters. This approach allows risk assessment to consider both the interactions between diseases and the individual's current health status, thereby improving the comprehensiveness and reliability of the assessment results. Furthermore, with a clear network structure, the influence relationships and degrees between diseases are very intuitive, and the resulting assessment data is highly interpretable. Compared to black-box models, this approach is more conducive to practical deployment and easier for healthcare professionals to understand the model's output.

[0040] Step S4: Generate individualized intervention recommendations based on the dynamic risk assessment results. A preferred method in this embodiment is: If the disease-related worsening risk assessment parameters If the value is greater than the corresponding preset threshold, an individual disease node is issued. The corresponding disease complications and exacerbation risk warning, this step mainly identifies the risk of synergistic deterioration of existing diseases.

[0041] If the disease-associated risk assessment parameters If the value is greater than the corresponding preset threshold, then a hidden node is emitted. The corresponding disease complication risk warning, this step is to identify potential disease risks that may occur in the future.

[0042] If individual patient's health risk If the value is greater than the corresponding preset threshold, an individual patient's health indicator risk warning will be issued to identify the risk of overall health deterioration.

[0043] In this step, on the one hand, by identifying the enhanced risk transmission between existing diseases through the disease-associated worsening risk assessment parameters, the worsening trend of current diseases under the synergistic effect of multiple diseases can be accurately located, thereby achieving precise monitoring and intervention of the impact of disease interactions and improving the targeting of health management. On the other hand, by assessing hidden nodes through disease-associated induced risk assessment parameters, potential diseases that have not yet occurred but have a development trend are included in the monitoring scope, enabling the system to identify the risk of emerging diseases in advance, significantly enhancing the foresight of health management. Furthermore, the degree of abnormality of multiple disease-related vital signs indicators is comprehensively quantified through the individual's overall health risk, reflecting the changing trend of the individual's health status.

[0044] In summary, step S4, through comprehensive coverage of current deterioration risks, potential precipitating risks, and overall health risks, makes health monitoring for people with multiple comorbidities more targeted, forward-looking, and comprehensive, thereby providing them with more accurate, continuous, and predictive health risk management methods.

[0045] Example 2 This embodiment provides a dynamic risk network modeling and real-time comprehensive risk assessment system for elderly individuals with multiple comorbidities, applied to the aforementioned dynamic risk network modeling and real-time comprehensive risk assessment method for elderly individuals with multiple comorbidities. (See reference...) Figure 2 ,include: The population-level baseline risk network construction module is used to build a population-level baseline risk network based on historical medical data, which reflects the average disease status of multiple patients.

[0046] The Individual Risk Network Construction Module is used to build an individual risk network for each individual patient, reflecting the individual patient's disease status.

[0047] The dynamic update module is used to periodically update the individual risk network.

[0048] The risk assessment module is used to conduct dynamic risk assessments based on population-level baseline risk networks and individual patient risk networks.

[0049] The results generation module is used to generate individualized intervention recommendations based on the results of dynamic risk assessment.

[0050] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly individuals with multiple comorbidities, characterized in that: Includes the following steps: A population-level baseline risk network is constructed based on historical medical data to reflect the average disease status of multiple patients. It includes disease nodes, disease parameters, directed edges, and baseline edge weights. Each disease node represents a disease type. Disease parameters that characterize disease-related vital signs are set on the disease nodes. The influence relationship between disease nodes is represented by directed edges. The baseline edge weights represent the strength of the influence relationship between upstream disease nodes and downstream disease nodes. An individual risk network is constructed for each individual patient to reflect their disease status. This network includes individual disease nodes, individual disease parameters, hidden nodes, directed edges, and risk edge weights. Each individual disease node represents a disease type possessed by the individual patient. Individual disease parameters are set on each individual disease node to represent disease-related vital signs. The directed edge connections between individual diseases are consistent with the population-level baseline risk network. Risk edge weights are used to indicate the probability that real-time data from upstream individual disease nodes will induce or cause the deterioration of disease parameters in downstream disease nodes. Hidden nodes are nodes that are not individual disease nodes but belong to disease nodes and are downstream nodes of the disease nodes corresponding to the individual disease nodes. The individual risk network is periodically updated dynamically, and an individual risk network is constructed based on the population-level baseline risk network and individual patients for dynamic risk assessment. Individualized intervention recommendations are generated based on the results of dynamic risk assessment.

2. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 1, characterized in that, The method for constructing the population-level baseline risk network is as follows: Access historical medical data and medical knowledge databases for multiple patients; Identify disease milestones based on historical medical data; Based on the average value of disease-related vital signs data of all patients in historical medical data, disease parameters are assigned to each disease node. The influence relationships between disease nodes are determined based on a medical knowledge database, and corresponding directed edges are established. The edge weights of each directed edge are calculated based on the historical medical data.

3. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 2, is characterized in that, Assuming the directed edge exists from disease node i to disease node j, the method for calculating the baseline edge weight based on the historical medical data is as follows: Calculate the probability parameter of coexistence of the diseases corresponding to disease node i and disease node j in the same patient based on historical medical data. and ratio ratio parameter : ； ; in, This represents the percentage of all patients with disease node i in the historical medical data who also have disease node j. This represents the number of patients in the historical medical data who have both disease node i and disease node j. This represents the number of patients in the historical medical data who have disease node i but do not have disease node j. This represents the number of patients in the historical medical data who do not have disease node i but do have disease node j. This represents the number of patients in the historical medical data who do not have disease node i and disease node j. The ratio between disease node i and disease node j. and These are the first threshold and the second threshold, respectively, with the second threshold being greater than the first threshold; Calculate the baseline edge weights of the directed edges from disease node i to disease node j in the population-level baseline risk network. : 。 4. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 1, characterized in that, The method for constructing the individual risk network is as follows: Obtain the population-level baseline risk network, retain the disease nodes corresponding to the diseases of individual patients and the downstream disease nodes of the disease nodes corresponding to the existing diseases, and delete other nodes; Traverse the remaining disease nodes. If an individual patient does not have the disease, define the disease node as a hidden node; otherwise, define it as an individual disease node. Based on the diagnostic data, the disease parameters of the individual disease node are modified to the disease-related vital signs data of the individual patient; Assume there exists a directed edge from individual disease node i to individual disease node j, and the risk edge weight is... The calculation method is as follows: ; ； in, Let be the baseline edge weight of the directed edge from disease node i to disease node j in the population-level baseline risk network, where disease node i' in the population-level baseline risk network and individual disease node i' in the individual risk network are of the same disease, and the value of i' is within the range of all disease nodes in the population-level baseline risk network. `clip` is used to... A function that clips values ​​to [0,1]. For intermediate parameters, It is an adjustment coefficient and is a positive number. This represents the deviation of the k-th disease parameter of individual disease node i from its corresponding normal value. It is the average value of the deviation of the k-th disease parameter of disease node i in the population-level baseline risk network from the corresponding normal value.

5. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 4, characterized in that, The method for dynamically updating the individual risk network is as follows: Obtain the most recent diagnostic data, update the disease parameters and the disease node, and re-apply according to the risk edge weights. Calculation method .

6. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 1, characterized in that, The method for dynamic risk assessment based on the population-level baseline risk network and the individual risk network is as follows: The disease-inducing risk of an individual patient is obtained based on the risk edge weight and the baseline edge weight; Individual patient health risk is obtained based on the disease parameters of the individual risk network.

7. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 6, is characterized in that, The method for obtaining the disease-inducing risk of an individual patient based on the risk edge weights and the baseline edge weights is as follows: Obtain the individual disease node i in the individual risk network, and obtain all its upstream nodes in the individual risk network. Calculate the disease association worsening risk assessment parameters for individual disease node i. : ； Among them, individual disease node b is an individual disease node. The upstream node, and These are individual disease node b to individual disease node b. The risk edge weights and their corresponding baseline edge weights in the population-level baseline risk network. For individual disease nodes The total number of upstream nodes; Obtain hidden nodes in the individual risk network And obtain all upstream nodes in the individual risk network, and calculate hidden nodes. Disease-associated risk assessment parameters : ； Among them, hidden nodes Corresponding individual disease nodes Individual disease nodes For individual disease nodes The upstream node, and Individual disease nodes To individual disease nodes The risk edge weights and their corresponding baseline edge weights in the population-level baseline risk network. For individual disease nodes The total number of upstream nodes.

8. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 7, is characterized in that, Individual patient health risk is obtained based on the disease parameters of the individual risk network. The method is as follows: ； ； in, This represents the health risk of the disease corresponding to individual disease node i. This represents the deviation of the k-th disease parameter of individual disease node i from its corresponding normal value. The total number of disease parameters representing individual disease node i. This represents the total number of individual disease nodes.

9. The method for dynamic risk network modeling and real-time comprehensive risk assessment for elderly people with multiple comorbidities as described in claim 8, characterized in that, The method for generating individualized intervention recommendations based on dynamic risk assessment results is as follows: If the disease-related worsening risk assessment parameters If the value is greater than the corresponding preset threshold, an individual disease node is issued. Warning of the risk of complications and worsening of the corresponding disease; If the disease-associated risk assessment parameters If the value is greater than the corresponding preset threshold, then a hidden node is emitted. Warning of the risk of complications from corresponding diseases; If individual patient's health risk If the value is greater than the corresponding preset threshold, an individual patient's health indicator risk warning will be issued.

10. A dynamic risk network modeling and real-time comprehensive risk assessment system for elderly people with multiple comorbidities, applied to the dynamic risk network modeling and real-time comprehensive risk assessment method for elderly people with multiple comorbidities as described in any one of claims 1-9, characterized in that, include: The population-level baseline risk network construction module is used to build a population-level baseline risk network based on historical medical data, which reflects the average disease status of multiple patients. The Individual Risk Network Construction Module is used to build an individual risk network for each individual patient to reflect their disease status. The dynamic update module is used to periodically update the individual risk network. The risk assessment module is used to conduct dynamic risk assessment by constructing individual risk networks based on population-level baseline risk networks and individual patients. The results generation module is used to generate individualized intervention recommendations based on the results of dynamic risk assessment.

Images