A method and system for individualized risk stratification for patients with coronary heart disease

Abstract

The present application relates to the technical field of medical information processing, and particularly relates to a method and system for individual risk stratification of coronary heart disease patients, which acquires multi-dimensional risk-related data of coronary heart disease patients in at least one preset time range, and performs time alignment and structured processing on the multi-dimensional risk-related data to construct a risk structure reflecting the correlation of multi-dimensional risk characteristics; performs evolution analysis on the risk structure in the time dimension to form a risk structure sequence arranged in chronological order; establishes an individual risk baseline based on the risk structure sequence, and determines the stability state of the risk structure and the risk structure instability path in combination with the change of the risk structure; further stratifies the patients according to the stability state and the instability path. When new multi-dimensional risk-related data is acquired, the risk structure and the time evolution result thereof are updated, and the risk stratification result is synchronously updated. The system is used to realize the above method.

Description

Technical Field

[0001] This invention relates to the field of medical information processing technology, and in particular to a method and system for individualized risk stratification of patients with coronary heart disease. Background Technology

[0002] Coronary artery disease (CAD) is a heart condition caused by insufficient blood supply to the coronary arteries, and its occurrence and development are closely related to multiple risk factors. In the health management of CAD patients, risk stratification is an important technical means to assess the patient's disease risk level and assist in the development of follow-up and intervention strategies. Individualized risk stratification refers to differentiated risk assessment based on the patient's own multidimensional risk-related information, so that the risk stratification results can reflect the risk differences between different patients and at different times for the same patient, thereby improving the pertinence and adaptability of risk assessment.

[0003] Existing methods for risk stratification of coronary artery disease typically rely on limited risk indicators collected at fixed time points, using pre-defined scoring rules or statistical models to classify patients into risk levels. However, in practice, these methods often treat different risk factors as independent and fail to reflect the overall evolution of a patient's risk status over time. As a patient's risk-related data is continuously updated, existing risk stratification results are difficult to adjust in a timely manner, leading to discrepancies between the risk stratification results and the patient's current risk status. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a personalized risk stratification method and system for patients with coronary heart disease, aiming to improve the problems of existing risk stratification methods that are difficult to reflect the evolution of patients' risk status over time and that the stratification results are difficult to adjust in a timely manner when risk-related data is updated.

[0005] In a first aspect, the present invention provides the following technical solution: a method for individualized risk stratification of patients with coronary heart disease, comprising: S1. Obtain multidimensional risk-related data of patients with coronary heart disease within at least one preset time range, and use the multidimensional risk-related data as input for subsequent risk analysis; S2. Process the multidimensional risk-related data and combine them according to preset risk categories to construct a risk structure that characterizes the risk composition relationship of an individual patient. S3. The risk structure is continuously characterized in the time dimension to form a risk structure sequence that reflects the change of risk structure over time; S4. Based on the risk structure of the corresponding baseline time period in the risk structure sequence, establish the individual risk baseline for the patient; S5. Analyze the changes in the risk structure based on the risk structure sequence to form change trajectory data reflecting the process of risk structure change; S6. Based on the change trajectory data and the individual risk baseline, determine the stability state of the risk structure, and determine the corresponding risk structure instability path when the risk structure is in an unstable state. S7. Based on the stability state of the risk structure and the instability path of the risk structure, perform individualized risk stratification for patients with coronary heart disease.

[0006] Preferably, in step S2, the risk structure construction step includes: The acquired multidimensional risk-related data are aligned according to a unified time index, and corresponding missing data are generated. The aligned multidimensional risk-related data is converted into a numerical sequence of risk features; Based on the risk feature sequence, the correlation metric between different risk features is calculated, and the correlation metric is organized into a risk structure in the form of a structural matrix.

[0007] Preferably, in step S3, the risk structure temporalization step includes: An initial time window is generated based on the sampling timestamps of the multidimensional risk-related data; The initial time window is adjusted based on the differences between the structure matrices within adjacent time windows. Within each adjusted time window, a risk structure sequence is formed, arranged in chronological order.

[0008] Preferably, in step S4, the individual risk baseline establishment step includes: Multiple structure matrices corresponding to the baseline time period are selected from the risk structure sequence; Robust statistical summarization processing is performed on the multiple structure matrices to form a baseline structure matrix; Reference range data corresponding to the baseline structure matrix is ​​generated based on the degree of dispersion of the plurality of structure matrices.

[0009] Preferably, in step S5, the risk structure disturbance analysis step includes: Generate structural difference sequences based on structural matrix differences between adjacent time windows; The structural difference sequence is used to identify change points to determine the time location of changes in the risk structure; Extract the structural matrix before and after the change in time and location, and calculate the matrix-level differences to form change trajectory data.

[0010] Preferably, in step S6, the step of determining the stability of the risk structure includes: The matrix-level differences in the trajectory data are quantified to obtain a stability assessment sequence. The stability assessment sequence is compared with the reference range data over time windows. The stability status of the risk structure is determined based on the comparison results.

[0011] Preferably, in step S6, the step of determining the risk structure instability path includes: When the risk structure is determined to be in an unstable state, the set of associated items that first exceed the preset change condition is identified based on the change of the associated metric value in the structure matrix. In subsequent time windows, a set of related items that satisfy the preset change conditions will be continuously extracted; The risk structure instability path is formed according to the chronological order of changes in the aforementioned set of related items.

[0012] Preferably, in step S7, the risk stratification determination step includes: The stability state of the risk structure is encoded, and the instability path of the risk structure is encoded. A hierarchical input code is generated based on the state code and path code; The risk stratification category corresponding to the coronary heart disease patient is determined based on the stratified input code.

[0013] Preferably, the dynamic update execution process includes: After acquiring new multidimensional risk-related data, the new multidimensional risk-related data will be incorporated into the risk-related data within the current time frame; The risk structure is reconstructed based on the risk-related data after the merger, and an updated risk structure sequence is formed. The stability state, instability path, and corresponding risk stratification categories of the risk structure are re-determined based on the updated risk structure sequence.

[0014] Secondly, the present invention provides the following technical solution: an individualized risk stratification system for patients with coronary heart disease, comprising: The data acquisition module is used to collect multidimensional risk-related data of patients with coronary heart disease within at least one preset time range; The risk structure construction module is used to perform time alignment on the multidimensional risk-related data and generate a missing label sequence to obtain a numerical risk feature sequence. Under the constraint of the missing label sequence, the module calculates the correlation metric between different risk features and organizes the correlation metric into structured data in the form of a structure matrix. The time processing module is used to generate time window boundaries based on the sampling timestamps of multidimensional risk-related data, and adjust the time window boundaries based on the differences between adjacent structure matrices, forming a risk structure sequence arranged in chronological order within the adjusted time window; The individual risk baseline establishment module is used to select multiple structure matrices corresponding to the baseline time period in the risk structure sequence, perform robust statistical summarization on the multiple structure matrices to generate a baseline structure matrix, and generate reference range data based on the dispersion of the multiple structure matrices. The disturbance analysis module is used to generate a structural difference sequence based on the structural matrix differences between adjacent time windows, identify change points in the structural difference sequence, extract the structural matrix of the time windows before and after the change points, and calculate matrix-level differences to form change trajectory data. The stability and instability path determination module is used to quantify the matrix-level differences in the change trajectory data and compare them with the reference range data window by window to determine the stability state of the risk structure. When the risk structure is in an unstable state, it identifies the set of related items based on the changes in the related metric values ​​in the structure matrix and forms instability path data. The risk stratification module is used to perform state encoding on the stability state of the risk structure and path encoding on the unstable path data, and to determine the risk stratification category corresponding to the coronary heart disease patient based on the encoding results.

[0015] The present invention has the following beneficial effects: 1. In this invention, by constructing a risk structure that reflects the correlation between multidimensional risk characteristics and performing evolutionary analysis on the risk structure in the time dimension, risk stratification is no longer based on isolated indicators or a single point in time, but is determined based on the overall state of the patient's risk structure and its change process. This enables the characterization of the intrinsic features of the patient's risk status evolving over time, providing a structured basis for individualized risk stratification.

[0016] 2. In this invention, by establishing an individualized risk baseline and corresponding reference range in the risk structure sequence, and comparing subsequent changes in the risk structure with the reference range, the stability of the risk structure is determined, and the basis for risk stratification is changed from a fixed threshold to a dynamic reference based on the patient's own historical structural characteristics, thereby enhancing the adaptability of the risk stratification results to individual differences.

[0017] 3. In this invention, when the risk structure is determined to be in an unstable state, an unstable path is formed by identifying the temporal sequence of changes in the correlation in the risk structure. The stable state and the unstable path are then uniformly encoded to achieve a comprehensive reflection of the risk evolution direction and change process of the risk stratification results. This allows the risk stratification results to be continuously updated with new data, supporting the continuous execution of the risk stratification method. Attached Figure Description

[0018] Figure 1 This is a flowchart of a method and system for individualized risk stratification of patients with coronary heart disease proposed in this invention; Figure 2 This is a system architecture diagram of an individualized risk stratification method and system for patients with coronary heart disease proposed in this invention. Detailed Implementation

[0019] The technical solutions in 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1: In the first embodiment of the present invention, the present invention provides a method and system for individualized risk stratification of patients with coronary heart disease, such as... Figure 1 As shown, it includes the following steps: S1. Obtain multidimensional risk-related data of patients with coronary heart disease within at least one preset time range, and use the multidimensional risk-related data as input for subsequent risk analysis.

[0021] Specifically, firstly, it is necessary to determine the time frame for risk analysis and use this time frame as a unified time benchmark to collect various risk-related data generated by patients within this time frame. The risk-related data comes from different data sources, and each data point is accompanied by a corresponding time stamp during acquisition. Then, the acquired raw risk-related data is preprocessed, converting non-numerical data into numerical representations and aligning data from different sampling frequencies according to the unified time benchmark. For data locations with missing values ​​during time alignment, corresponding missing markers are generated to distinguish between true values ​​and missing states.

[0022] After completing time alignment and missing data labeling, the multidimensional risk-related data is organized into a time series format in chronological order, so that each time index corresponds to a set of risk-related feature values. This time series can be represented as: ; in: Indicates a time index; Indicates time index Risk-related data vectors at the location; Indicates the first Risk-related characteristics in time index The possible values ​​of ; This indicates the number of risk-related characteristics.

[0023] Finally, this time-series multidimensional risk-related data is stored as input data for subsequent risk analysis.

[0024] This step generates multidimensional risk-related data input organized chronologically under a unified time index, providing a consistent data foundation for subsequent risk structure construction and risk evolution analysis.

[0025] S2. Process the multidimensional risk-related data and combine them according to the preset risk categories to construct a risk structure that characterizes the risk composition relationship of individual patients. Furthermore, in step S2, the risk structure construction step includes: The acquired multidimensional risk-related data are aligned according to a unified time index, and corresponding missing data are generated. The aligned multidimensional risk-related data is converted into a numerical sequence of risk features; The correlation measure between different risk features is calculated based on the risk feature sequence, and the correlation measure is organized into a risk structure in the form of a structure matrix.

[0026] Specifically, firstly, the multidimensional risk-related data is time-aligned according to a unified time index, so that different risk features correspond to each other under the same time index. For data locations missing at a certain time index, a corresponding missing marker is generated to indicate the completeness status of the risk feature values ​​under that time index. Then, the multidimensional risk-related data after time alignment and missing marker processing is converted into a numerical risk feature sequence, so that each risk feature forms a continuous data representation in the time dimension.

[0027] After obtaining the risk feature sequence, a correlation metric is calculated between different risk features to characterize their relationship over time. The correlation metric between different risk features can be expressed as: ; in: and They represent the first The and the first Risk characteristics in time index The numerical sequence below; This represents a metric function used to calculate the correlation between risk features, and the function is preferably Euclidean distance or cosine similarity; Indicates risk characteristics Risk characteristics The correlation metric between them.

[0028] Then, the correlation metrics between the various risk characteristics are organized according to the risk characteristic index to form a risk structure in the form of a structure matrix, which is used to characterize the overall combination relationship between the individual risk characteristics of patients.

[0029] By aligning multidimensional risk-related data under a unified time index and calculating correlation metrics between risk features, the relationships between risk features are no longer independent but are expressed holistically in a structured form. This transforms the multidimensional risk status of patients within the same time frame into a risk structure representation that can be used for subsequent evolutionary analysis.

[0030] This step generates a risk structure in the form of a structural matrix that reflects the interrelationships of multidimensional risk characteristics, providing a unified data representation basis for subsequent time-evolution-based risk structure analysis and stability determination.

[0031] S3. Continuously characterize the risk structure in the time dimension to form a risk structure sequence that reflects the changes in risk structure over time; Furthermore, in step S3, the risk structure temporalization step includes: An initial time window is generated based on the sampling timestamps of multidimensional risk-related data; The initial time window boundary is adjusted based on the difference between the structure matrices in adjacent time windows; Within each adjusted time window, a risk structure sequence is formed, arranged in chronological order.

[0032] Specifically, firstly, initial time windows are generated based on the sampling timestamps of multidimensional risk-related data, ensuring that each time window covers the structural matrix formed within its corresponding time range, thereby segmenting and organizing the risk structure in the time dimension. Then, differences are calculated between the structural matrices formed within adjacent time windows, and the boundaries of the initial time windows are adjusted based on these differences, so that periods of concentrated structural change form independent time windows.

[0033] The difference in the structure matrix corresponding to adjacent time windows can be expressed as: ; in: Indicates the first The structure matrix formed within a time window; This represents the structure matrix formed within the adjacent previous time window; Represents the matrix norm used to measure the differences in structural matrices; This represents the structural difference between adjacent time windows.

[0034] After adjusting the time window boundaries, the corresponding structure matrices are arranged in chronological order within each adjusted time window, thus forming a risk structure sequence that describes the evolution of risk structure over time.

[0035] By introducing structural matrix differences as the basis for adjusting the time window, the time segmentation not only depends on the sampling time distribution, but also reflects the changes in the risk structure itself, thereby obtaining a structural sequence representation that is more in line with the characteristics of risk evolution in the time dimension.

[0036] This step organizes the discrete structure matrix into a time-sequential sequence of risk structures, providing a continuous data foundation for subsequent time-evolution-based stability analysis of risk structures and identification of unstable paths.

[0037] S4. Based on the risk structure of the corresponding baseline time period in the risk structure sequence, establish the individual risk baseline for patients; Furthermore, in step S4, the individual risk baseline establishment step includes: Select multiple structure matrices corresponding to the baseline time period from the risk structure sequence; Robust statistical summarization is performed on multiple structure matrices to form a baseline structure matrix; Reference range data corresponding to the baseline structure matrix is ​​generated based on the degree of dispersion of multiple structure matrices.

[0038] Specifically, firstly, a baseline time period for baseline construction is determined within the risk structure sequence, and multiple structure matrices are selected from this baseline time period to characterize the risk structure distribution of patients under relatively stable conditions. Then, robust statistical summarization is performed on the selected structure matrices to reduce the impact of abnormal structures on baseline construction, resulting in a baseline structure matrix characterizing the individual risk structure center state of each patient.

[0039] The baseline structure matrix can be represented as: ; in: This represents the structure matrix selected within the baseline time period; This indicates an operation used for robust statistical summarization of multiple structure matrices; This represents the resulting baseline structure matrix; This indicates the number of structure matrices used for baseline construction.

[0040] After obtaining the baseline structure matrix, reference range data is generated based on the dispersion of multiple selected structure matrices relative to the baseline structure matrix. This data is used to characterize the normal fluctuation range of the risk structure within the benchmark time period. The dispersion can be calculated from the difference between the structure matrix and the baseline structure matrix, thus forming the reference range data corresponding to the baseline structure matrix.

[0041] By selecting a baseline time period in the risk structure sequence and performing robust statistical summarization on the structure matrix, the baseline structure matrix reflects the typical state of the patient's own risk structure. At the same time, the degree of dispersion characterizes the allowable range of structural changes under the baseline state, providing an individualized reference basis for subsequent risk structure stability assessment.

[0042] This step establishes a risk baseline that reflects the central state of an individual patient's risk structure and its normal fluctuation range, providing a clear comparative basis for subsequent risk structure change analysis and stability assessment.

[0043] S5. Analyze the changes in risk structure based on the risk structure sequence to generate change trajectory data that reflects the process of risk structure change; Furthermore, in step S5, the risk structure disturbance analysis step includes: Generate structural difference sequences based on structural matrix differences between adjacent time windows; Identify change points in the structural difference sequence to determine the temporal location of changes in the risk structure; Extract the structural matrix before and after the change in time and location, and calculate the matrix-level differences to form change trajectory data.

[0044] Specifically, after obtaining the risk structure sequence arranged in chronological order, the changes in the risk structure over time are analyzed to identify the time points at which the risk structure changes and to characterize its change process.

[0045] First, based on the differences between structural matrices within adjacent time windows, the changes in each adjacent structural matrix are quantified, and the quantification results are arranged in chronological order to form a structural difference sequence. Then, change points are identified within the structural difference sequence to determine the temporal locations of changes in risk structures, thus identifying the time indices within the structural change set as the change time locations.

[0046] After determining the time position of the change, the structural matrix before and after the corresponding time position is extracted, and matrix-level difference calculation is performed on the structural matrix. The differences of the structural matrix that change over time are organized in chronological order to form change trajectory data to describe the process of risk structural change.

[0047] Through the above steps, the changes in risk structure within adjacent time windows are quantified in the form of structural difference sequences. Based on this, the time position of the risk structure changes is identified, thereby forming trajectory data reflecting the change process of risk structure over time. This provides a clear data foundation for subsequent determination of risk structure stability and identification of instability paths based on the change trajectory.

[0048] S6. Based on the change trajectory data and individual risk baseline, determine the stability state of the risk structure, and determine the corresponding risk structure instability path when the risk structure is in an unstable state. Furthermore, in step S6, the step of determining the stability of the risk structure includes: Quantification of matrix-level differences in the trajectory data is performed to obtain a stability assessment sequence. The stability assessment sequence was compared with the reference range data over time windows. The stability status of the risk structure is determined based on the comparison results.

[0049] Specifically, after generating trajectory data reflecting the changes in the risk structure, the stability of the risk structure over time is determined.

[0050] First, the matrix-level differences reflecting the degree of structural matrix change in the change trajectory data are quantified. These time-varying matrix-level differences are then organized into a stability assessment sequence arranged in chronological order to describe the distribution of the intensity of risk structure changes within continuous time windows. Next, the stability assessment sequence is compared with the reference range data formed in the previous steps, window-by-window, to determine whether the structural changes within each time window are within the reference range.

[0051] After completing the time-window comparison, the stability state of the risk structure within the corresponding time range is determined based on the comparison results, so that the stable or unstable state of the risk structure in the time dimension is clearly identified.

[0052] This step transforms the change trajectory data into a stability assessment sequence that can be used for time window-level comparisons. Combined with individualized reference ranges, the change state of the risk structure is determined, thereby achieving an objective determination of the stability state of the risk structure and providing a basis for stability judgment for subsequent identification of risk structure instability paths.

[0053] Furthermore, in step S6, the step of determining the risk structure instability path includes: When the risk structure is determined to be in an unstable state, the set of related items that first exceed the preset change conditions are identified based on the changes in the related metrics in the structure matrix. In subsequent time windows, continuously extract the set of related items that meet the preset change conditions; The risk structure instability path is formed by the chronological order in which the set of related items changes.

[0054] Specifically, after determining the stability state of the risk structure, when the risk structure is determined to be in an unstable state, the process of the risk structure evolving from stable to unstable is analyzed in order to determine the instability path of the risk structure.

[0055] First, based on the changes in each correlation metric in the structure matrix over time, the magnitude of these changes is calculated, and correlations with magnitudes exceeding preset change conditions are identified as changing correlations, thus forming a set of correlations within the corresponding time window. Then, in subsequent time windows, the changes in correlation metrics in the structure matrix are continuously analyzed, and a set of correlations satisfying preset change conditions is extracted within each time window, ensuring that continuously changing correlations in the risk structure are recorded window by window.

[0056] After extracting the sets of related items for each time window, the sets of related items are arranged in chronological order of when they first meet the preset change conditions, thus forming the risk structure instability path used to describe the evolution of the risk structure from a stable state to an unstable state.

[0057] This step organizes the changes in the relationships within the risk structure into an instability path in chronological order, clearly depicting the evolution of the risk structure from a stable to an unstable state. This provides a structured input basis for subsequent individualized risk stratification based on the instability path. S7. Individualized risk stratification of patients with coronary heart disease based on the stability state of the risk structure and the instability path of the risk structure; Further, in step S7, the risk stratification determination step includes: State coding is performed on the stability state of the risk structure, and path coding is performed on the instability path of the risk structure; Generate a hierarchical input code based on the state code and path code; The risk stratification category for patients with coronary heart disease is determined based on the stratified input code.

[0058] Specifically, after obtaining the stability state and instability path of the risk structure, risk stratification is determined for patients with coronary artery disease. First, the stability state of the risk structure is encoded, converting it into a discretized state identifier for subsequent processing. Simultaneously, the instability path of the risk structure is encoded, converting the sequence of structural changes reflected in the instability path into corresponding path identifiers, thus representing both the stability and evolutionary characteristics of the risk structure in coded form. Then, a stratified input code is generated based on the state and path codes, allowing the stability state and instability path to form a combined representation within the same coding space.

[0059] After generating the stratified input code, the risk stratification category corresponding to the coronary heart disease patient is determined based on the stratified input code, so that the patient is classified into the risk stratification corresponding to the stability state and evolution path of his / her risk structure.

[0060] This step combines the stability state and instability path of the risk structure in a unified coding form, and determines the risk stratification category accordingly. This allows the risk stratification results to be directly based on the stability characteristics and evolution process of the risk structure, thereby completing the determination of individualized risk stratification for patients with coronary heart disease.

[0061] Furthermore, the dynamic update execution process includes: After acquiring new multidimensional risk-related data, the new multidimensional risk-related data will be incorporated into the risk-related data within the current time frame; The risk structure is reconstructed based on the risk-related data after the merger, and an updated risk structure sequence is formed. The stability state, instability path, and corresponding risk stratification categories of the risk structure are re-determined based on the updated risk structure sequence.

[0062] Specifically, after determining the risk stratification based on existing multidimensional risk-related data, the above method is dynamically updated and executed when new multidimensional risk-related data is acquired.

[0063] First, the newly acquired multidimensional risk-related data is merged into the existing risk-related data within the current time frame. The merged data is then time-aligned according to established time indexing rules, ensuring that the new data and existing data form a continuous dataset under a unified time benchmark. Next, the risk structure is reconstructed based on the merged risk-related data, creating an updated risk structure sequence along the time dimension. This ensures that the risk structure reflects the latest changes following the introduction of new data.

[0064] After the updated risk structure sequence is formed, the stability state of the risk structure, the instability path of the risk structure, and the corresponding risk stratification category are re-determined based on the risk structure sequence, so that the risk stratification results are adjusted synchronously with the data update.

[0065] Through the above dynamic update execution process, after introducing new multidimensional risk-related data, the risk structure and its temporal evolution characteristics can be continuously updated, and the stability state, instability path and risk stratification results of the risk structure can be re-determined, thereby realizing the continuous execution of the risk stratification method in the time dimension.

[0066] Example 2: In the long-term follow-up and dynamic monitoring of patients with coronary heart disease, the patient's relevant risk factors continuously change over time. Existing systems typically assess risk based on fixed time points or single indicators, making it difficult to reflect the evolution of the patient's risk status over time. Furthermore, as new data is continuously generated, existing risk stratification results cannot be effectively updated, leading to discrepancies between the risk stratification results and the patient's current actual condition. To address these issues, this invention provides an individualized risk stratification system for patients with coronary heart disease, the structure of which is as follows: Figure 2 As shown. The specific implementation process of this system is as follows: The data acquisition module collects multidimensional risk-related data of patients with coronary heart disease within at least one preset time range, and adds time stamps to the collected data so that data from different sources have a basis for alignment in the time dimension.

[0067] The risk structure construction module performs time alignment processing on the collected multidimensional risk-related data and generates missing data markers. It then converts the processed data into a numerical risk feature sequence. Based on this, it calculates the correlation metrics between different risk features and organizes the correlation metrics into a risk structure in the form of a structure matrix.

[0068] The temporal processing module generates an initial time window based on the sampling timestamps of the multidimensional risk-related data after forming the structure matrix. It then adjusts the boundaries of the time window based on the differences between the structure matrices in adjacent time windows, thereby forming a risk structure sequence arranged in chronological order in the time dimension.

[0069] The individual risk baseline establishment module selects multiple structural matrices corresponding to the baseline time period in the risk structure sequence, performs robust statistical summarization on the selected structural matrices to form the baseline structural matrix, and generates reference range data corresponding to the baseline structural matrix based on the dispersion of the structural matrix.

[0070] After obtaining the risk structure sequence and baseline structure, the disturbance analysis module generates a structural difference sequence based on the differences between the structure matrices of adjacent time windows, and identifies the change points in the structural difference sequence to determine the time position of the risk structure change, thereby forming change trajectory data reflecting the process of risk structure change.

[0071] The stability and instability path determination module quantifies the matrix-level differences in the change trajectory data and compares the quantification results with the reference range data in a time window to determine the stability state of the risk structure. When the risk structure is determined to be in an unstable state, it identifies the set of related items based on the changes in the correlation metrics in the structure matrix and forms the instability path of the risk structure according to the time sequence of the changes in the set of related items.

[0072] The risk stratification module encodes the stability state of the risk structure and the path of instability of the risk structure. It generates a stratification input code based on the state code and the path code, and determines the risk stratification category corresponding to the coronary heart disease patient based on the stratification input code.

[0073] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for individualized risk stratification of patients with coronary heart disease, characterized in that, include: S1. Obtain multidimensional risk-related data of patients with coronary heart disease within at least one preset time range, and use the multidimensional risk-related data as input for subsequent risk analysis; S2. Process the multidimensional risk-related data and combine them according to preset risk categories to construct a risk structure that characterizes the risk composition relationship of an individual patient. S3. The risk structure is continuously characterized in the time dimension to form a risk structure sequence that reflects the change of risk structure over time; S4. Based on the risk structure of the corresponding baseline time period in the risk structure sequence, establish the individual risk baseline for the patient; S5. Analyze the changes in the risk structure based on the risk structure sequence to form change trajectory data reflecting the process of risk structure change; S6. Based on the change trajectory data and the individual risk baseline, determine the stability state of the risk structure, and determine the corresponding risk structure instability path when the risk structure is in an unstable state. S7. Based on the stability state of the risk structure and the instability path of the risk structure, perform individualized risk stratification for patients with coronary heart disease.

2. The method for individualized risk stratification of patients with coronary heart disease according to claim 1, characterized in that, In step S2, the risk structure construction step includes: The acquired multidimensional risk-related data are aligned according to a unified time index, and corresponding missing data are generated. The aligned multidimensional risk-related data is converted into a numerical sequence of risk features; Based on the risk feature sequence, the correlation metric between different risk features is calculated, and the correlation metric is organized into a risk structure in the form of a structural matrix.

3. The individualized risk stratification method for patients with coronary heart disease according to claim 1, characterized in that, In step S3, the risk structure temporalization step includes: An initial time window is generated based on the sampling timestamps of the multidimensional risk-related data; The initial time window is adjusted based on the differences between the structure matrices within adjacent time windows. Within each adjusted time window, a risk structure sequence is formed, arranged in chronological order.

4. The individualized risk stratification method for patients with coronary heart disease according to claim 1, characterized in that, In step S4, the individual risk baseline establishment step includes: Multiple structure matrices corresponding to the baseline time period are selected from the risk structure sequence; Robust statistical summarization processing is performed on the multiple structure matrices to form a baseline structure matrix; Reference range data corresponding to the baseline structure matrix is ​​generated based on the degree of dispersion of the plurality of structure matrices.

5. The method for individualized risk stratification of patients with coronary heart disease according to claim 1, characterized in that, In step S5, the risk structure disturbance analysis step includes: Generate structural difference sequences based on structural matrix differences between adjacent time windows; The structural difference sequence is used to identify change points to determine the time location of changes in the risk structure; Extract the structural matrix before and after the change in time and location, and calculate the matrix-level differences to form change trajectory data.

6. The method for individualized risk stratification of patients with coronary heart disease according to claim 1, characterized in that, In step S6, the step of determining the stability of the risk structure includes: The matrix-level differences in the trajectory data are quantified to obtain a stability assessment sequence. The stability assessment sequence is compared with the reference range data over time windows. The stability status of the risk structure is determined based on the comparison results.

7. The method for individualized risk stratification of patients with coronary heart disease according to claim 1, characterized in that, In step S6, the step of determining the risk structure instability path includes: When the risk structure is determined to be in an unstable state, the set of associated items that first exceed the preset change condition is identified based on the change of the associated metric value in the structure matrix. In subsequent time windows, a set of related items that satisfy the preset change conditions will be continuously extracted; The risk structure instability path is formed according to the chronological order of changes in the aforementioned set of related items.

8. The method for individualized risk stratification of patients with coronary heart disease according to claim 1, characterized in that, In step S7, the risk stratification determination step includes: The stability state of the risk structure is encoded, and the instability path of the risk structure is encoded. A hierarchical input code is generated based on the state code and path code; The risk stratification category corresponding to the coronary heart disease patient is determined based on the stratified input code.

9. The individualized risk stratification method for patients with coronary heart disease according to claim 1, characterized in that, The dynamic update execution process includes: After acquiring new multidimensional risk-related data, the new multidimensional risk-related data will be incorporated into the risk-related data within the current time frame; The risk structure is reconstructed based on the risk-related data after the merger, and an updated risk structure sequence is formed. The stability state, instability path, and corresponding risk stratification categories of the risk structure are re-determined based on the updated risk structure sequence.

10. A personalized risk stratification system for patients with coronary heart disease, characterized in that, include: The data acquisition module is used to collect multidimensional risk-related data of patients with coronary heart disease within at least one preset time range; The risk structure construction module is used to perform time alignment on the multidimensional risk-related data and generate a missing label sequence to obtain a numerical risk feature sequence. Under the constraint of the missing label sequence, the module calculates the correlation metric between different risk features and organizes the correlation metric into structured data in the form of a structure matrix. The time processing module is used to generate time window boundaries based on the sampling timestamps of multidimensional risk-related data, and adjust the time window boundaries based on the differences between adjacent structure matrices, forming a risk structure sequence arranged in chronological order within the adjusted time window; The individual risk baseline establishment module is used to select multiple structure matrices corresponding to the baseline time period in the risk structure sequence, perform robust statistical summarization on the multiple structure matrices to generate a baseline structure matrix, and generate reference range data based on the dispersion of the multiple structure matrices. The disturbance analysis module is used to generate a structural difference sequence based on the structural matrix differences between adjacent time windows, identify change points in the structural difference sequence, extract the structural matrix of the time windows before and after the change points, and calculate matrix-level differences to form change trajectory data. The stability and instability path determination module is used to quantify the matrix-level differences in the change trajectory data and compare them with the reference range data window by window to determine the stability state of the risk structure. When the risk structure is in an unstable state, it identifies the set of related items based on the changes in the related metric values ​​in the structure matrix and forms instability path data. The risk stratification module is used to perform state encoding on the stability state of the risk structure and path encoding on the unstable path data, and to determine the risk stratification category corresponding to the coronary heart disease patient based on the encoding results.

Images