Interpretable machine learning model for cardiovascular disease risk prediction: a feature decomposition-based study.

Area of Science:

• Computational biology
• Medical informatics
• Machine learning in healthcare

Background:

• Cardiovascular disease (CVD) poses a significant global health challenge.
• Early prediction and identification of CVD risk factors are crucial for prevention.
• Machine learning (ML) offers promising tools for developing predictive models.

Purpose of the Study:

• To construct and validate machine learning models for predicting cardiovascular disease (CVD) risk.
• To evaluate the performance of a novel feature decomposition-based deep learning (FDDL) model.
• To identify key predictors of CVD using model interpretability techniques.

Main Methods:

• Utilized a large dataset of 68,205 respondents from Kaggle.
• Developed and tested a feature decomposition-based deep learning (FDDL) model.
• Compared FDDL against six other ML models and employed SHAP for interpretation.

Main Results:

• The FDDL model achieved high predictive performance: 75.52% accuracy, 78.14% precision, 71.68% recall, F1 score of 0.7522, and AUC-ROC of 0.7643.
• Diastolic blood pressure, cholesterol, systolic blood pressure, and age were identified as critical predictors.
• The Logistic Regression (LR) model showed the weakest performance.

Conclusions:

• An effective ML model for CVD risk prediction was developed.
• The model can assist clinicians in identifying high-risk individuals.
• Provides a basis for personalized preventive healthcare strategies for cardiovascular disease.