Effective Heart Disease Prediction Using Machine Learning with Five Classifiers

Abstract

Heart disease is among the most difficult diseases, and it affects a lot of individuals all over the world. Early and accurate heart illness identification is vital in the world of medicine, particularly in the field of cardiology. Lower mortality rates could arise from the prevention or treatment of heart disease in the early stages. It takes more accuracy, perfection, and correctness to diagnose and predict heart-related disorders. Finding the greatest ML algorithm that can effectively predict cardiac disease is the system's major goal. We employed five hybrid classifiers, including the Decision Tree (DT), the Random Forest (RF), the Gradient Boosting Method (GBM), the Support Vector Machine (SVM), and the k-nearest neighbor algorithm (KNN). We have utilized the Univariate feature selection technique to choose key features. Moreover, we calculate F1 Score (F1), Precision (PRE), and Accuracy (ACC). The findings revealed that, when Univariate is taken into account, the RF classification algorithm achieves a comparably greater accuracy of roughly 98.31% than others. Thus, we discovered a relatively basic machine learning approach that may be utilized to create highly accurate heart disease prediction. In order to determine which attributes are more important in the model results, the chosen 08 features are also utilized to examine the model results for "interpretability".