Heart disease has affected people for centuries and is still a leading cause of death, so properly designed early diagnosis systems are vital. In this study, we use five fuzzy ensemble models to help predict a patient’s risk of heart attack. All three models were developed using Kaggle datasets that focus on clinical, demographic and lifestyle traits. Healthcare applications find fuzzy logic useful, since it handles fuzzy and uncertain data which commonly exists in healthcare. The Adaptive Fuzzy Random Forest model had higher accuracy, precision, recall and F1-score than its ability to handle membership functions adaptively. The F1-score attained was 0.7875 and the area under the ROC curve was 0.8768 on the dataset Heart Attack Prediction in Indonesia. The F1-score of 0.7888 and AUC-ROC of 0.8787 for Type-2 Fuzzy Random Forest show that it easily handles challenging class imbalance while maintaining an acceptable level of certainty. Using Fuzzy Weighted Random Forest together with SMOTE significantly increased recall and precision when the data was imbalanced. These results offer practical value since fuzzy ensemble methods make it possible to predict the risk of cardiovascular disease and should be useful when added to AI healthcare systems. More study will concentrate on improving speed and comparing these models with real medical information.