Heart Arrhythmia Detection and Classification: A Comparative Study Using Deep Learning Models

The irregular functioning of heartbeats known as "Heart Arrhythmia" may lead to heart palpitations, blood clots, and even a heart stroke. The ECG test is one of the primary clinical tests that are utilized to detect heart abnormalities due to its noninvasive nature. However, this method is an extremely time-consuming process due to variations in ECG signals. The main aim of this study is to automate this manual process as computer-aided detection can diagnose with more precision and accuracy. This research study is a comparative study that detects and classifies arrhythmia using various deep learning models that is a one-dimensional convolutional neural network, two-dimensional convolutional neural network (2D-CNN), 2D-CNN with long short-term memory network, and in addition to this, models are combined using ensemble learning to develop a classifier. These classifiers help discriminate signs of arrhythmia disease. The idea is implemented on ECG Heartbeat Categorization data, derived from the MIT-BIH arrhythmia dataset. The utilization of deep learning-based methods helps to achieve promising results and show significant improvements as compared to baseline methods. This study would benefit the medical experts in early arrhythmia diagnosis as faster detection can save more human lives.