Machine Learning Model for Predicting Coronary Heart Disease Risk: Development and Validation Using Insights From a...

Area of Science:

• Cardiovascular Disease Research
• Machine Learning in Healthcare
• Biostatistics and Epidemiology

Background:

• Coronary heart disease (CHD) is a leading global cause of death, necessitating improved risk assessment and prevention strategies.
• Traditional models for CHD risk assessment have limitations due to their reliance on predefined variables.
• Machine learning (ML) offers advanced analytical capabilities for complex, high-dimensional data to uncover novel CHD predictors.

Purpose of the Study:

• To evaluate the predictive power of established and novel risk factors for CHD incidence using ML techniques.
• To compare the performance of various ML models in predicting CHD.
• To identify key predictors of CHD and understand their contribution using model interpretation methods.

Main Methods:

• A cohort of 7260 participants (aged 30-84) from Suita City, Japan, was followed for an average of 15 years for cardiovascular events.
• Five ML models (logistic regression, random forest, SVM, XGBoost, LightGBM) were employed to predict CHD incidence.
• Model performance was assessed using metrics like accuracy, AUC, and calibration, with Shapley Additive Explanations (SHAPs) used for risk factor interpretation.

Main Results:

• The Random Forest model achieved the highest predictive performance (accuracy 0.73, AUC 0.73).
• Key predictors included intima-media thickness, blood pressure, lipid profiles, and estimated glomerular filtration rate.
• Novel significant contributors to CHD risk were identified as lower calcium levels, elevated white blood cell counts, and body fat percentage, with a protective effect noted in women.

Conclusions:

• A robust ML model, interpreted by SHAP, effectively predicts CHD incidence by considering multifactorial risks.
• The findings underscore the importance of integrating novel biomarkers and considering gender-specific factors in CHD risk assessment.
• This data-driven approach can support healthcare professionals in developing targeted prevention strategies for cardiovascular health.