Predictive Modelling for Heart Disease: Evaluating Machine Learning Techniques

Heart disease (HD) attitudes as the major terminal illness worldwide; therefore, health professionals require efficient predictive systems to identify diseases in their early stages. A clinical dataset containing 303 patients along with features like cholesterol levels, age, and exercise-induced angina serves as the basis for analysing three machine learning (ML) replicas namely Logistic Regression (LR) and Random Forest (RF) and Support Vector Machine (SVM). The evaluation of the reproductions occurred through examination of accuracy alongside Precision and Recall while measuring F1-Score laterally with ROC-AUC and displaying Matthews Correlation Coefficient (MCC) and Cohen’s Kappa scores. The evaluation indicates RF as the leading model because it produces 94% accuracy along with an ROC-AUC score of 0.97 for superior prediction results. The performance of SVM matches closely to LR but achieves 90% accuracy while LR reaches 77% accuracy. The discovered effectiveness of ensemble learning methods demonstrates RF as a remarkable model for heart ailment prediction improvement.