Risk prediction system and risk prediction device for stable coronary heart disease
By integrating multi-dimensional risk factor data and building a comprehensive risk prediction model, the problems of insufficient accuracy of stable coronary heart disease risk assessment and limited complex data processing capabilities in the prior art are solved, and more efficient and reliable risk assessment is achieved.
Patent Information
- Application Number
- CN202510244073.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-03
- Publication Date
- 2025-06-13
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has problems such as insufficient accuracy, strong subjectivity and limited processing capabilities for complex data in the risk assessment of stable coronary heart disease.
By integrating traditional risk factors, imaging data, biomarker data and individualized lifestyle data, a comprehensive risk prediction model is built, and a machine learning algorithm and integrated learning method are used, and a regularization method is combined to build a risk prediction model.
It improves the accuracy and reliability of stable coronary heart disease risk assessment, enhances the processing ability of complex data, and provides more scientific risk prediction results.
Smart Images
Figure CN120148859A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of coronary heart disease risk prediction, and specifically provides a risk prediction system and a risk prediction device for stable coronary heart disease. Background Art
[0002] Stable coronary heart disease refers to a type of coronary heart disease in which the clinical symptoms and myocardial ischemia status of patients are relatively stable within a certain period of time. Compared with unstable coronary heart disease, the risk assessment of stable coronary heart disease is more complex because its symptoms are not typical and are easily overlooked. However, patients with stable coronary heart disease may still experience acute cardiovascular events under certain inducing factors (such as strenuous exercise, emotional excitement, etc.).
[0003] Traditional methods for coronary heart disease risk assessment mainly rely on the experience of clinicians and the detection of some simple biomarkers. Although these methods can provide risk assessment to a certain extent, they often have defects such as insufficient accuracy, strong subjectivity, and limited ability to process complex data. Summary of the Invention
[0004] Aiming at the deficiencies of the prior art, the present invention provides a risk prediction system and a risk prediction device for stable coronary heart disease. By integrating multi-dimensional risk factor data, including traditional risk factors, imaging data, biomarker data, and individualized lifestyle data, a comprehensive risk prediction model is constructed to improve the accuracy and reliability of risk assessment, thereby solving the above technical problems.
[0005] To achieve the above object, the present invention provides the following technical solution: A risk prediction system for stable coronary heart disease, comprising:
[0006] A risk data acquisition module, configured to collect multi-dimensional risk factor data, where the multi-dimensional risk factor data includes: traditional risk factor data, imaging data, biomarker data, and individualized lifestyle data;
[0007] A risk feature fusion module, configured to receive the preprocessed multi-dimensional risk factor data from the risk data acquisition module, and perform feature extraction and fusion on different types of data;
[0008] A risk prediction module, configured to construct a risk prediction model according to the fusion data obtained after being processed by the risk feature fusion module and in combination with a machine learning algorithm. The data expression formula of the risk prediction model is as follows:
[0009]
[0010] Wherein, R is the risk score of the patient developing stable coronary heart disease, α is the constant term (intercept) of the model, F iis the i-th risk factor (such as traditional risk factors) after feature selection, W i is the weight coefficient related to the i-th risk factor, reflecting the impact degree of this factor on the risk, F j and F k are the interaction terms of two risk factors, used to capture the interaction between different factors, W n+j is the weight coefficient related to the interaction term;
[0011] A risk assessment module, used to evaluate and interpret the prediction results generated by the risk prediction model, and the prediction results include: risk level classification, individual risk factor analysis, and risk trend analysis.
[0012] Preferably, the traditional risk factor data includes but is not limited to age, gender, family history, smoking history, hypertension, diabetes, and blood lipid data, the imaging data includes but is not limited to coronary CT angiography, echocardiogram, and radionuclide myocardial perfusion imaging, the biomarker data includes but is not limited to cardiac troponin, B-type natriuretic peptide or N-terminal pro-B-type natriuretic peptide, and high-sensitivity C-reactive protein, and the individual lifestyle data includes but is not limited to eating habits, exercise habits, psychological stress level, and sleep quality data.
[0013] Preferably, the system further includes a risk data preprocessing module, and the risk data preprocessing module is used to perform data cleaning, standardization, and feature engineering processing on the multi-dimensional risk factor data.
[0014] Preferably, the risk feature fusion module uses at least one of principal component analysis (PCA), linear discriminant analysis (LDA), or deep neural network for feature extraction and fusion.
[0015] Preferably, the risk feature fusion module specifically includes:
[0016] A data receiving unit, used to receive the multi-dimensional risk factor data preprocessed by the risk data acquisition module, and the multi-dimensional risk factor data is transmitted in a structured data format, including numerical data and categorical data;
[0017] A feature extraction unit, used to perform feature extraction on the received different types of data respectively. For numerical data, methods such as standardization, normalization, or principal component analysis are used to extract features, and for categorical data, methods such as one-hot encoding, label encoding, or target encoding are used to extract features.
[0018] Preferably, the risk feature fusion module further includes:
[0019] A data fusion unit for fusing the features of different types of data after extraction, and the fusion method includes but is not limited to weighted fusion, splicing fusion or a fusion method based on deep learning;
[0020] A data standardization unit for standardizing the selected feature data.
[0021] Preferably, in the risk prediction model, the interaction term F j and F k , when calculating, uses the product of different risk factors as the interaction feature to capture the non-linear interaction relationship between risk factors, and the weight coefficient W n+j is obtained through model training and learning.
[0022] Preferably, when constructing the risk prediction model using machine learning algorithms, the risk prediction module adopts an ensemble learning method to fuse the prediction results of multiple base models, and the fusion method includes at least one of voting method, averaging method or weighted averaging method.
[0023] Preferably, during the process of constructing the risk prediction model, the risk prediction module adopts a regularization method, and the regularization term includes at least one of L1 regularization, L2 regularization or elastic net regularization, and determines the value of the regularization parameter through cross-validation.
[0024] The present invention also provides a risk prediction device for stable coronary heart disease, the prediction device includes a memory and one or more processors, the memory stores the executable code of the prediction system, and the processor is used to run the executable code stored in the memory.
[0025] Compared with the prior art, the present invention provides a risk prediction system and a risk prediction device for stable coronary heart disease, having the following beneficial effects:
[0026] The risk prediction system and device for stable coronary heart disease provided by the present invention integrate multi-dimensional data such as traditional risk factors, imaging, biomarkers and individualized lifestyle, comprehensively reflect the health status of patients, and improve the accuracy of risk assessment; through data preprocessing and feature fusion, ensure data quality, provide a high-quality basis for model construction, the risk prediction model considers the interaction of factors, combines ensemble learning and regularization methods, and improves the stability and accuracy of prediction; the risk assessment module can perform risk level classification, individualized analysis and trend prediction, provide a scientific basis for clinical decision-making and patient management, and help improve the prevention, treatment and prognosis of patients with stable coronary heart disease, and overall improve the scientificity and practicality of coronary heart disease risk prediction. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] Figure 1 is the logical block diagram of the system of the present invention. Detailed implementation manners
[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0029] Please refer to Figure 1 , a risk prediction system for stable coronary heart disease provided by the present invention, includes:
[0030] A risk data acquisition module, configured to collect multi-dimensional risk factor data, and the multi-dimensional risk factor data includes: traditional risk factor data, imaging data, biomarker data, and individualized lifestyle data;
[0031] A risk feature fusion module, configured to receive the multi-dimensional risk factor data preprocessed by the risk data acquisition module, and perform feature extraction and fusion on different types of data;
[0032] A risk prediction module, configured to construct a risk prediction model according to the fusion data obtained after being processed by the risk feature fusion module and in combination with a machine learning algorithm. The data expression formula of the risk prediction model is as follows:
[0033]
[0034] Wherein, R is the risk score of the patient having stable coronary heart disease, α is the constant term (intercept) of the model, F i is the i-th risk factor (such as a traditional risk factor) after feature selection, W i is the weight coefficient related to the i-th risk factor, reflecting the influence degree of this factor on the risk, F j and F k are the interaction terms of two risk factors, used to capture the interaction between different factors, and W n+j is the weight coefficient related to the interaction term;
[0035] A risk assessment module, configured to evaluate and interpret the prediction results generated by the risk prediction model, and the prediction results include: risk level classification, individualized risk factor analysis, and risk trend analysis.
[0036] Furthermore, traditional risk factor data includes but is not limited to age, gender, family history, smoking history, hypertension, diabetes, and lipid data. Imaging data includes but is not limited to coronary CT angiography, echocardiogram, and radionuclide myocardial perfusion imaging. Biomarker data includes but is not limited to cardiac troponin, B-type natriuretic peptide or N-terminal pro-B-type natriuretic peptide, and high-sensitivity C-reactive protein. Individualized lifestyle data includes but is not limited to eating habits, exercise habits, psychological stress levels, and sleep quality data.
[0037] By setting up a risk data acquisition module to collect the above data, these data cover multiple aspects such as individual physiological characteristics, disease-related indicators, and living habits, and can comprehensively reflect the health status of patients. Traditional risk factor data helps to evaluate the disease risk from a basic level; imaging data can visually present the structure and function status of the heart blood vessels, etc., and assist in accurate judgment; biomarker data can provide disease clues at the microscopic level; individualized lifestyle data can analyze risks from the perspective of daily behaviors. Combining these multi-dimensional data can more accurately construct a risk prediction model for stable coronary heart disease and improve the accuracy and reliability of risk assessment.
[0038] Furthermore, the system also includes a risk data preprocessing module, which is used to perform data cleaning, standardization, and feature engineering processing on multi-dimensional risk factor data.
[0039] Furthermore, the risk feature fusion module uses at least one of principal component analysis (PCA), linear discriminant analysis (LDA), or deep neural network for feature extraction and fusion.
[0040] When the risk data preprocessing module processes multi-dimensional risk factor data, it mainly performs data cleaning, standardization, and feature engineering processing; first of all, data cleaning aims to remove noise and outliers in the data to ensure the accuracy and consistency of the data; this includes dealing with missing values, for example, solving the problem of data missing through methods such as deletion, filling, or interpolation.
[0041] Furthermore, the risk feature fusion module specifically includes:
[0042] A data receiving unit, which is used to receive the multi-dimensional risk factor data preprocessed by the risk data acquisition module. The multi-dimensional risk factor data is transmitted in a structured data format, including numerical data and categorical data;
[0043] A feature extraction unit, which is used to perform feature extraction on the received different types of data respectively. For numerical data, methods such as standardization, normalization, or principal component analysis are used to extract features, and for categorical data, methods such as one-hot encoding, label encoding, or target encoding are used to extract features;
[0044] A data fusion unit, which is used to fuse the features of different types of data after extraction, and the fusion methods include but are not limited to weighted fusion, splicing fusion or fusion methods based on deep learning;
[0045] A data standardization unit, which is used to perform standardization processing on the selected feature data.
[0046] The process of the risk feature fusion module fusing data is as follows: the data receiving unit receives the preprocessed multi-dimensional risk factor data containing numerical and categorical data; then, the feature extraction unit uses means such as standardization, normalization or principal component analysis for numerical data to extract key features, and for categorical data, uses methods such as one-hot encoding, label encoding or target encoding for feature extraction; then, the data fusion unit uses weighted fusion, splicing fusion or fusion methods based on deep learning to integrate the features of different types of data. Weighted fusion sums the weights according to the feature importance, splicing fusion directly splices the feature vectors, and the deep learning fusion method obtains the fusion features by learning complex relationships through a neural network; finally, the data standardization unit performs standardization processing on the fused feature data to ensure that the data is on the same scale, providing a high-quality, stable and comprehensive data basis for the construction of the subsequent risk prediction model.
[0047] Further, the interaction term F in the risk prediction model j and F k , when calculating, uses the product of different risk factors as the interaction feature to capture the non-linear interaction relationship between risk factors, and the weight coefficient W of the interaction term n+j is obtained through model training.
[0048] Further, when the risk prediction module constructs a risk prediction model using a machine learning algorithm, it adopts an ensemble learning method to fuse the prediction results of multiple base models, and the fusion methods include at least one of voting method, averaging method or weighted averaging method.
[0049] Further, the risk prediction module adopts a regularization method in the process of constructing the risk prediction model. The regularization terms include at least one of L1 regularization, L2 regularization or elastic net regularization, and the value of the regularization parameter is determined through cross-validation.
[0050] The risk prediction model calculates the risk score of a patient having stable coronary heart disease by combining multiple risk factors and their interactions. First, the model uses the risk factors after feature selection (such as traditional risk factors, imaging data, biomarker data and individualized lifestyle data), and assigns corresponding weight coefficients (W i ), and these weight coefficients reflect the degree of influence of each factor on the risk; then, the model introduces the interaction term of the risk factor (Fj and F k ), and these interaction terms capture the non - linear interaction relationships between factors through the product of different risk factors, and assign weight coefficients (W n+j ) to each interaction term. These weight coefficients are learned through model training; the data expression of the risk prediction model is:
[0051]
[0052] Among them, R is the risk score, and α is the constant term (intercept) of the model. During the model training process, an ensemble learning method (such as voting method, averaging method or weighted averaging method) is adopted to fuse the prediction results of multiple base models to improve the stability and accuracy of prediction; at the same time, the model adopts a regularization method (such as L1, L2 or elastic net regularization) and determines the regularization parameter through cross - validation to prevent overfitting and optimize the model performance; finally, the model generates the final risk score based on these calculation results, and conducts risk level classification, individual risk factor analysis and risk trend analysis, so as to provide comprehensive risk prediction results.
[0053] The risk assessment module deeply analyzes the results of the risk prediction model through multi - dimensional assessment methods. In terms of risk level classification, it accurately classifies patients into different risk levels, such as low risk, medium risk and high risk, etc., based on the risk score obtained from the risk prediction model and referring to the pre - set risk level thresholds, so that medical staff and patients can intuitively understand the severity of the condition; during individual risk factor analysis, it deeply explores the contribution of each risk factor to individual risk, and by analyzing the weight coefficients of each risk factor, etc., clarifies which factors have a greater impact on the occurrence and development of stable coronary heart disease in this patient, thus providing a key basis for personalized treatment and intervention; in terms of risk trend analysis, it combines the data of multiple evaluations of the patient, observes the change trend of the risk score over time, and predicts the development direction of the condition, so as to adjust the treatment plan in a timely manner; by setting this module, complex prediction results can be transformed into clear and practical information, which helps clinicians make more scientific diagnostic decisions, improves patients' awareness of their own conditions, and thus more effectively prevents and manages stable coronary heart disease and improves the prognosis of patients.
[0054] The present invention also provides a risk prediction device for stable coronary heart disease. The prediction device includes a memory and one or more processors. The memory stores the executable code of the prediction system, and the processor is used to run the executable code stored in the memory.
[0055] The memory stores the executable code of the prediction system, providing the basic support for the entire risk prediction system and ensuring the proper preservation of the data and algorithms required for the system operation. The processor, as the computing core of the device, is responsible for running the executable code in the memory and can quickly process the multi-dimensional risk factor data collected by the risk data acquisition module, including traditional risk factors, imaging data, biomarker data, and individualized lifestyle data, etc. By executing the code, the processor calls the risk feature fusion module to extract and fuse the features of the data, constructs a risk prediction model using the risk prediction module and calculates the risk results, and finally evaluates and interprets the results with the help of the risk assessment module.
[0056] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A risk prediction system for stable coronary heart disease, characterized by: include, A risk data acquisition module, used to collect multi-dimensional risk factor data, wherein the multi-dimensional risk factor data includes: traditional risk factor data, imaging data, biomarker data and individualized lifestyle data; A risk feature fusion module is used to receive the multi-dimensional risk factor data pre-processed by the risk data acquisition module and perform feature extraction and fusion on different types of data; The risk prediction module is used to construct a risk prediction model based on the fusion data obtained after processing by the risk feature fusion module and combined with the machine learning algorithm. The data expression of the risk prediction model is as follows: Among them, R is the patient's risk score for stable coronary heart disease, α is the constant term (intercept) of the model, and F i is the ith risk factor after feature selection (such as traditional risk factors), W i is the weight coefficient associated with the ith risk factor, reflecting the impact of this factor on the risk. j and F k is the interaction term of two risk factors, which is used to capture the interaction between different factors. n+j is the weight coefficient associated with the interaction term; The risk assessment module is used to evaluate and interpret the prediction results generated by the risk prediction model, which include: risk level classification, individualized risk factor analysis and risk trend analysis.
2. A risk prediction system for stable coronary heart disease according to claim 1, characterized in that: The traditional risk factor data include but are not limited to age, gender, family history, smoking history, hypertension, diabetes and blood lipid data; the imaging data include but are not limited to coronary artery CT angiography, echocardiography and radionuclide myocardial perfusion imaging; the biomarker data include but are not limited to cardiac troponin, B-type natriuretic peptide or N-terminal pro-B-type natriuretic peptide and high-sensitivity C-reactive protein; the personalized lifestyle data include but are not limited to eating habits, exercise habits, psychological stress level and sleep quality data.
3. A risk prediction system for stable coronary heart disease according to claim 1, characterized in that: The system also includes a risk data preprocessing module, which is used to perform data cleaning, standardization and feature engineering on multi-dimensional risk factor data.
4. The risk prediction system for stable coronary heart disease according to claim 1, characterized in that: The risk feature fusion module uses at least one of principal component analysis (PCA), linear discriminant analysis (LDA) or deep neural network to extract and fuse features.
5. The risk prediction system for stable coronary heart disease according to claim 1, characterized in that: The risk feature fusion module specifically includes: A data receiving unit, used to receive the multi-dimensional risk factor data pre-processed by the risk data acquisition module, wherein the multi-dimensional risk factor data is transmitted in a structured data format, including numerical data and categorical data; The feature extraction unit is used to extract features from different types of received data. For numerical data, standardization, normalization or principal component analysis is used to extract features. For categorical data, one-hot encoding, label encoding or target encoding is used to extract features.
6. A risk prediction system for stable coronary heart disease according to claim 1, characterized in that: The risk feature fusion module also includes: A data fusion unit, used to fuse the extracted features of different types of data, wherein the fusion method includes but is not limited to weighted fusion, splicing fusion or a fusion method based on deep learning; The data standardization unit is used to standardize the selected feature data.
7. The risk prediction system for stable coronary heart disease according to claim 1, characterized in that: The interaction term F in the risk prediction model j and F k , the product of different risk factors is used as the interaction feature in the calculation to capture the nonlinear interaction relationship between risk factors, and the weight coefficient of the interaction term W n+j Learned through model training.
8. The risk prediction system for stable coronary heart disease according to claim 1, characterized in that: When the risk prediction module uses a machine learning algorithm to construct a risk prediction model, it adopts an integrated learning method to fuse the prediction results of multiple base models, and the fusion method includes at least one of a voting method, an averaging method or a weighted average method.
9. The risk prediction system for stable coronary heart disease according to claim 1, characterized in that: The risk prediction module adopts a regularization method in the process of constructing a risk prediction model. The regularization term includes at least one of L1 regularization, L2 regularization or elastic network regularization, and the value of the regularization parameter is determined by cross-validation.
10. A risk prediction device for stable coronary heart disease, applicable to a risk prediction system for stable coronary heart disease according to any one of claims 1 to 9, characterized in that: The prediction device includes a memory and one or more processors, the memory stores executable codes of the prediction system, and the processor is used to run the executable codes stored in the memory.