Advanced Deep Learning Techniques for Accurate Prediction of Heart Diseases Using Electrocardiogram Signal Analysis

Abstract

Early detection of heart disease is critical to the patient's survival. An electrocardiogram (ECG) is a test that analyses heartbeat variations. ECG is a test that checks on how your heartbeats vary. Various cardiac diseases can be detected by the deviation of signals from the typical sinus rhythm as well as from mere anomalies. The ECG signal carries minor amplitude variation, thus can cause errors as it may be difficult to make a diagnosis on the cardiac conditions. The only way to preserve the human lives is by the very accurate recognition method. The ECG signals are utilized in an appropriate and accurate way for classifying and predicting the heart diseases through a proposed study in this study. In the study, Convolutional Neural networks (CNN), Visual Geometry Group (VGG) and Logistic Regression (LR) were employed to predict the heart diseases; the results proved out robust and finally, ensemble approaches were developed based on the combination of CNN, VGG, LR with Bidirectional Recurrent Neural Network(BRNN), Gated Recurrent unit, and Long Short term memory which are used to predict heart diseases and performance of each as discussed.