Ensemble Deep Learning Model to Enhance Heart Disease Prediction

Heart disease prediction (HDP) is one of the important medical issues that can lower health risks. Although several studies have been done on deep learning (DL) and machine learning (ML) algorithms, the accuracy of HDP needs to be improved. This study aimed to enhance HDP accuracy using a proposed stacking ensemble-deep learning (SE-DL) model. The SE-DL ensembled three pre-trained DL algorithms: recurrent neural network (RNN), long short-term memory (LSTM), and gated recurrent units (GRU). Furthermore, it also integrated with two stacking ML models: logistic regression (LR) and support vector machine (SVM), to improve the performance of HDP. Evaluating the effectiveness of the proposed model used four performance metrics: accuracy, recall, precision, and F1-score. The experimental results showed that the SE-DL model performed better than single DL and ML algorithms. When applied to the datasets without feature selection, it was better than with several feature selection methods.