Classification of Arrhythmia Diseases using Hybrid Novel Models dependent on CNNs and Long Shorter Term Memory with Particle Swarm Optimization Algorithm

Abstract

In this paper, the classification of arrhythmia diseases using hybrid novel models dependent on CNNs and Long Shorter Term Memory with Particle Swarm Optimization Algorithm is presented along with the simulation results. An electro-cardiogram, viz., ECG serves as a non- invasing type of diagnostic tool for cardiac’s arrhythmias (CA’s). The accurate identifications of CAs relies on effective classification methods, which have traditionally employed diverse mathematical & computational strategies. In these studies, we present a novel computational based models utilizing the particle swarm’s optimization [PSO] algorithms, convolutional neural nets [CNN], and long short-term memory (LSTM) for the classifications of 6 CA class sourced from the MIT based BIH Arrhythmias Datasets [MITDB]. The primary objective of the PSOs are to optimizing the hyper-parameters defining the layered based architectures of the CNN, aiming to enhance accuracies while minimizing categorical based cross entropial errors [CE]. The outcomes underscore the reliability of the proposed model, signifying an innovative approach that eliminates the need for manual hyperparameter selection in the layered architectures of the CNNs based on LSTM. This research explores the Classification of Arrhythmia Diseases through the innovative integration of Convolutional Neural Nets (CNN’s) & the Long Short Term Memory [LSTM] network within a Hybrid Models. The study leverages the optimization capabilities of the Particle Swarm Optimization (PSO) Algorithm to enhance the model's performance. The synergistic combination of these technologies aims to improve the accuracy and efficiency of automated arrhythmia classification, contributing to advancements in medical diagnostics and patient care.