XGBoost for Heart Disease Prediction Achieving High Accuracy with Robust Machine Learning Techniques

The comprehensive dataset on heart disease presented in this study consists of 1190 cases with 11 shared characteristics from five well-known datasets: Cleveland, Hungarian, Switzerland, Long Beach, Virginia, and Statlog. Because of this, it is the biggest dataset of its kind for studies on coronary artery disease (CAD). To aid in early detection, a robust machine learning model that could reliably forecast cardiac illness needed to be developed. To eliminate null values and divide the dataset into an 80:20 train-test ratio, we employed exploratory data analysis. To ensure that the characteristics were consistent, we also employed conventional scaling. Logistic Regression, Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbours, Gradient Boosting, AdaBoost, and XGBoost were the eight machine learning methods that we examined. Optimized using grid search with 5-fold cross-validation, XGBoost performed the best with test accuracy of 0.966, precision of 0.967, and recall of 0.966. Three false positives and one false negative could be distinguished by it. The approach may be helpful in clinical settings, as evidenced by its high recall for positive cases (0.986). By providing us with a new dataset and an effective predictive model, this work advances the diagnosis of CAD. This makes it possible to identify and treat CAD earlier.