Coronary Artery Disease Prediction Using Machine Learning

Machine learning (ML) is being widely used to help diagnose diseases and track how they progress. In this project, we used different ML techniques on the Z-Alizadeh Sani public UCI dataset, which focuses on Coronary Artery Disease (CAD). The main aim was to predict CAD using methods like Logistic Regression (LR), Random Forest (RF), CatBoost, XGBoost, and LightGBM. While there are other systems for predicting CAD, many struggle with low accuracy or handling missing data. To tackle this, we used t-tests to select important features and PCA to reduce the dataset's dimensions because it had a small sample size. We tested our models using 10-fold CV, which gave us an average accuracy of over 86.02%. The models also showed precision and Fl-scores close to 90%. Among them, Random Forest performed the best, with an AUC of 87%. Overall, our approach showed promising results and performed better than previous studies.