Heart Disease Analysis Using Machine Learning

Cardiovascular diseases remain the leading cause of mortality worldwide. According to WHO (2023), cardiovascular diseases accounted for 17.9 million deaths globally, representing 32% of all deaths. This study proposes a machine learning–based heart disease prediction and analysis system integrated into a desktop application with role-based dashboards. The system includes role-based authentication and a Tkinter interface designed for usability. A pilot usability study is planned to quantify effectiveness. The proposed framework leverages multiple classifiers (Logistic Regression, Decision Tree, Random Forest, and K-Nearest Neighbors) for predictive analysis, evaluated through accuracy, ROC-AUC, calibration, and interpretability methods (SHAP). Results demonstrate that ensemble approaches, particularly Random Forest, achieve superior predictive performance. The system demonstrates potential as a research-oriented decision-support prototype, with future work focusing on external validation and clinical deployment readiness.