Prediction and Feature Importance Analysis for Heart Failure using Machine Learning Techniques

Abstract

Heart failure is a clinically complex syndrome that affects millions of people globally and posts a heavy strain on healthcare systems because of its high rates of morbidity and mortality. To manage and treat it effectively, early detection and accurate prediction are essential. Machine learning algorithms present a viable technique. This research attempts to forecast heart failure under specific circumstances, using a heart disease dataset from kaggle.com that is accessible to the public. Five machine learning models: Random Forest (RF), Decision Tree (DT), Logistic Regression (LR), Support Vector Machine (SVM), and K-Nearest Neighbor (KNN), were used as they show strong performance in applications pertaining to machine. The performance of five models was evaluated by accuracy, precision, recall and f1-score with 11 symptoms. From the calculation, Random Forest shows better performance than the remaining models with an accuracy of 88% and ST_Slope_Up is tested to have the greatest impact on the prediction.