Design and Implementation of Deep Learning Model for Atrial Fibrillation Classification using ECG Signals

Abstract

Electrocardiograms (ECGs), which are an essential diagnostic tool, are required to be performed in the normal course of clinical practise in order to evaluate cardiac arrhythmias. Convolutional neural network framework is suggested for use in this method, which makes use of deep learning to carry out automatic ECG arrhythmia diagnosis by classifying patient ECGs into the proper cardiac states. The prior training for this network was done using a standard signal data set. The primary objective of this approach is to provide a basic, reliable, and easily implemented deep learning algorithm for the categorization of the two separate cardiac category scenarios that have been selected. The findings demonstrated that a conventional back propagation neural network used in cascade with transferred deep learning classification was able to accomplish exceptionally high levels of performance. The primary objective of this research is to develop an efficient classification system that can forecast the severity of a patient's sleep apnea, as well as to improve classification accuracy and reduce the number of incorrect classifications.