Ensemble Deep Learning Models for Accurate Prediction of Cardiovascular Disease Risk: A Comparative Analysis

Abstract

A leading cause of death globally is cardiovascular disease (CVD). Early CVD detection is essential for successful treatment and complication prevention. Convolutional neural network (CNN), Recurrent neural networks (RNN), bidirectional recurrent neural networks (BiRNN), deep neural networks (DNN), and an ensemble model has all been used in this study's deep learning-based approach for CVD prediction. With a test size of 20%, suggested models were trained and assessed on a dataset of 303 patients. The models were assessed using a variety of criteria, including recall, sensitivity, specificity, F1-score, accuracy, and precision. The ensemble model achieved best performance, with 99% accuracy, 100% precision, 100% recall, 0.97 F1-score, 1.0 sensitivity, and 0.99 specificity. The training and validation loss vs. epoch graph for each model was also analysed to assess the model's performance. Findings from this research suggest that the proposed machine learning-based approach can effectively predict CVD, with the ensemble model outperforming individual models. The use of such models can aid in the early detection and prevention of CVD, improving patient outcomes. Future work can focus on evaluating the proposed models on a larger dataset and incorporating additional clinical variables.