Prediction of Heart Failure Disease Using Ensemble Model

Abstract

Heart failure remains one of the leading causes of morbidity and mortality worldwide, necessitating the development of accurate and efficient predictive systems for early diagnosis and prevention. This study presents a Heart Failure Predictive Model using an Ensemble Learning Approach, designed to enhance diagnostic accuracy and reliability through the combination of multiple machine learning algorithms. Clinical datasets were obtained from the UCI Machine Learning Repository and Kaggle, containing key patient attributes such as age, blood pressure, ejection fraction, and serum creatinine levels. Data preprocessing involved cleaning, normalization, and the use of the SMOTEENN technique to correct class imbalance. Relevant features were selected using the Lion Optimization Algorithm (LOA) to improve model performance and reduce computational complexity. Five base classifiers, Artificial Neural Network (ANN), Decision Tree (C4.5), Logistic Regression (LR), K-Nearest Neighbor (KNN), and Support Vector Machine (SVM) were individually trained and later combined using ensemble techniques such as Voting, AdaBoost, and Gradient Boosting. Experimental results demonstrated that ensemble methods significantly outperformed individual classifiers, achieving up to 99.75% accuracy using the boosting ensemble model. Performance metrics, including accuracy, precision, recall, F1-score, and Cohen’s kappa, confirmed the robustness and reliability of the model. The study concluded that ensemble learning provides a powerful, data-driven approach for early heart failure prediction, enabling timely clinical interventions and improved patient outcomes. The developed model serves as a foundation for integrating intelligent diagnostic systems into healthcare environments, contributing to the advancement of predictive medicine and the practical application of artificial intelligence in medical diagnostics.