ANALYSIS OF FUZZY RANDOM FOREST AND IT’S VARIANTS FOR HEART ATTACK ASSESSMENT

Abstract

This thesis examines the application of fuzzy ensemble methods, in this case Fuzzy Random Forests, to classify problems in medicine and specifically heart disease prediction. This research examines thoroughly the application of Fuzzy Random Forests in medicine from diabetes and asthma to liver disease, breast cancer, cholera, heart disease, and dentistry in dealing with uncertainty and complicated data. The latter section discusses how the five fuzzy ensemble models—Type-1, Type-2, Fuzzy Weighted Random Forest, Fuzzy Decision Forest, and Adaptive Fuzzy Random Forest—are performed on three publicly used datasets to predict heart attack. Results validate the improved performance of Adaptive Fuzzy Random Forest in terms of accuracy, precision, recall, and F1-score because of the adaptive control of membership functions. Fuzzy Weighted Random Forest performed well in dealing with imbalanced datasets, thus substantiating the practical validity of fuzzy logic in medical practice. The importance of fuzzy ensemble techniques in developing strong AI-based systems for medical diagnosis is emphasized through this thesis, and it offers directions for future enhancement in real clinical practice and computational efficiency.