An explainable machine learning framework for cardiovascular risk prediction using structured health data.

Area of Science:

• Cardiology
• Artificial Intelligence
• Machine Learning

Background:

• Cardiovascular disease (CVD) remains a leading global cause of death.
• Machine learning (ML) models are increasingly used for CVD risk prediction.
• Interpretability challenges hinder clinical adoption of many ML models.

Purpose of the Study:

• To introduce an interpretable ML framework for cardiovascular risk prediction.
• To enhance the transparency and clinical utility of ML models in healthcare.
• To identify key predictors of cardiovascular risk through explainable AI.

Main Methods:

• Utilized a cardiovascular dataset of approximately 70,000 patient records.
• Developed and evaluated Logistic Regression, Random Forest, and Gradient Boosting models using 5-fold cross-validation.
• Applied SHAP (Shapley Additive Explanations) for global and local feature interpretability.

Main Results:

• Ensemble-based ML models demonstrated superior predictive performance.
• Gradient Boosting achieved the highest Area Under the ROC Curve (AUC) at 0.794, closely followed by a Voting Ensemble model (0.793).
• All models significantly outperformed the baseline Logistic Regression (AUC 0.773).

Conclusions:

• Explainable AI techniques, particularly SHAP, successfully identified age, blood pressure, cholesterol, and weight as critical predictors.
• The proposed interpretable ML framework enhances transparency in cardiovascular risk prediction.
• This approach fosters trust and facilitates clinical decision-making for predictive healthcare models.