A Different View on Artificial Intelligence Applications for Cardiac Arrhythmia Detection and Classification

Ischemic heart disease and stroke are the world's highest killers. These diseases have continued the foremost cause of death globally in the last 15 years. An electrocardiogram (ECG or EKG) is a record of the electrical activity of the heart over a period of time, represented by one-dimensional data. In this paper, different methods such as Wavelet Transform will be applied to the ECG signal to increase accuracy on arrhythmia detection. The main arrhythmia disorders studied in this work were the following: atrial premature and supraventricular beat, the fusion of ventricular and normal beat, isolated QRS-like artifact, ventricular escape beat, and premature ventricular contraction. The annotation on each sample was made by certified cardiologists and each database was uploaded on Physionet (Research Resource for Complex Physiologic Signals). The input data used was 2D images instead of classical time-series data to demonstrate that the presented system can be used to identify with success different arrhythmias directly on images. In other words, a similar algorithm can be used in mobile devices such as smartphones. The test accuracy obtained demonstrates the efficiency of the system that could be applied to an electrocardiogram to easily detect any specific arrhythmia disorders.