Machine Learning Models Analysis using Multi-Factors for Heart Attack Risk Prediction

Abstract

In human population heart attacks is a leading cause of dying in world, so the need of accurate and timely risk assessment strategies is critically required. This comparative study presents an analysis of heart attack risk prediction by applying different machine learning methods such as Logistic Regression, Decision Tree, Random Forest etc. These methods were trained the model by using dataset of clinical features and then tested. The prediction of heart attack show based on various parameters such as accuracy, precision, recall, F1 score, training time, ROC-AUC, and interpretability. The prediction results indicates that XGBoost and Random Forest with selected top features provide higher predictive accuracy score of 98.49% and 99.01%. Decision trees provide faster processing and neural networks model give complex relationships. These models comparatively less stable and require greater resources. This study shows that the application of different machine learning techniques with appropriate feature selection as an effective solution for prediction of heart attack risk.