Predictive Modeling of Heart Failure Using Health Parameters and Machine Learning Techniques

Abstract

This study conducts a comprehensive analysis of machine learning models' potential in predicting heart failure using a dataset compiled from multiple sources across various locations. Through data preprocessing and analysis, significant correlations were identified between lifestyle characteristics and heart failure incidence. Several machine learning models, including Logistic Regression, Support Vector Machine, Random Forest, K-nearest neighbors, Extra trees, Gradient Boosting, and CatBoost, were developed, trained, and evaluated using performance metrics such as accuracy, feature importance, confusion matrix, and the ROC curve. The Random Forest model exhibited superior performance, emphasizing its robustness and effectiveness in heart failure prediction. This research underscores the significance of applying machine learning to enhance predictive accuracy and provides key insights for future applications in clinical decision support systems, suggesting directions for further research in expanding the models to encompass a broader range of cardiovascular conditions according to individual lifestyle.

Keywords: Heart Failure Prediction, Machine Learning Models, Lifestyle Characteristics, Clinical Decision Support Systems.

DOI: 10.7176/RHSS/14-6-01

Publication date: June 30th 2024