ECG signal processing using 1-D Convolutional Neural Network for Congestive Heart Failure Identification

Abstract

Heart disease is one of the leading causes of death in the world. Congestive Heart Failure (CHF) is one type of heart disease that needs attention. CHF is a condition in which the heart cannot pump blood adequately throughout the body. This disease usually affects patients over the age of 60 years. An EKG can be used to diagnose this condition. However, doctors need to diagnose manually, namely, reading the ECG signal directly. Therefore, this study aims to create a system that can diagnose CHF automatically using the 1D convolutional neural network (CNN) method. This CNN 1D method uses normalization as preprocessing, three hidden layers with 16 output channels, a fully connected layer, and sigmoid activation. The research dataset comes from MIT-BIH and BIDMC. Based on this study, 100% accuracy results were obtained with recall, precision, and 1 F1-Score, respectively, so this study can assist medical staff in identifying CHF conditions and providing appropriate therapy to patients.