Classification of Cardiac Arrhythmias using Recurrent Neural Network and selected features

Abstract

The most frequent kind of heart ailment is cardiac arrhythmia (also known as a tachycardia). The computer-based decision-making method is quite beneficial in the analysis of the Electrocardiogram (ECG) signal and the categorization of CAs, among other things. This research describes an automated categorization of CA's that combines chosen aspects of the ECG signal with a Bidirectional Long Short-Term Memory (BLSTM) network, which is described in detail elsewhere. The linear and non-linear components of the ECG data were extracted and input to two BLSTM networks, which were then coupled together in a fully connected layer. BLSTM networks are the most extensively used recurrent neural networks for evaluating sequential data and are also the most widely used recurrent neural networks. All of the characteristics of the segmented heartbeats are retrieved. The five main forms of CAs are discussed in detail. Normal Beat (N), Left Bundle Branch Block (L), Right Bundle Branch Block (R), Premature Ventricular Contraction (V), and Paced Beat are the five kinds of heartbeats (Q). The findings demonstrate that the BLSTM model, which incorporates both linear and nonlinear characteristics, achieves the maximum accuracy in the classification task at hand.