Detection and Classification of Arrhythmias by Deploying Deep Learning Models

Abstract

Abstract Arrhythmias can be detected using an ECG signal, which is an important tool in the healthcare industry. ECG overall variation trends, original variation features, and their relative positions are used to classify arrhythmias according to sphere knowledge and large-scale data analysis. They haven’t been fully explored by being styles. CNN and hybrid CNN-LSTM models are used to address this problem. A LSTM and CNN are used to separate the ECG’s overall variation trends and its unique features. In this project the implemented models are CNN and Hybrid LSTM models to check which model is better in identifying the arrythmias based on the ACC, SEN, and SPE scores. The Accuracy of the CNN model is 74.4 percent, respectively, while the Hybrid-CNN LSTM scores are 83.5 on the MIT-BIH arrhythmias dataset.