Prompt identification of cardiovascular disease can greatly reduce global mortality. We propose a comprehensive AI pipeline that combines Support Vector Machines, Random Forests, and XGBoost to estimate heart disease risk using 13 routine clinical features. Evaluating a cohort of 303 patients, our optimized XGBoost model achieved 94.7% accuracy, 95.2% recall, and 94.1% specificity. Explainability analyses highlight chest pain category, maximum heart rate, and exercise-induced ST depression as the top predictors. This framework offers both robust performance and clear interpretability, empowering clinicians with actionable insights.
Sarika Shinde (Wed,) studied this question.