A Deep Neural Network heartbeat classification approach for arrhythmia detection

Digital electrocardiogram (ECG) analysis acts as a crucial role in the clinical ECG; it is associated with high prevalence, high mortality rates, and sustained healthcare costs. The most popular ECG data and the robust deep learning algorithm made it possible to enhance the precision and scalability of electronic ECG interpretation. Still, there is no thorough evaluation of an end-to-end deep learning method used in ECG analysis. In this work, we extend a deep neural network (DNN) to classify rhythm classes using 91,232 single-lead ECGs from 53,549 patients who used a single-lead ambulatory ECG monitoring device. We classify eight rhythm classes using single-lead ECGs from 53,549 patients who used a single-lead ambulatory ECG monitoring device. When validated against an independent test dataset annotated by a consensus committee of board-certified practicing cardiologists, the potential of applying Deep Neural Network (DNN) approaches to the automatic prediction of arrhythmia has been largely overlooked thus far. This study addresses this important gap by presenting a DNN model that accurately identifies arrhythmia depending on one raw electrocardiogram (ECG) heartbeat only, also collate prevailing algorithms based on Heart Rate divergence. We trained and tested the model using ECG datasets, comprising 380000 heartbeats, to achieve 100% arrhythmia prediction accuracy. Notably, the model also identifies those heartbeat sequences and ECG's morphological characteristics, which are class-discriminative and thus prominent for arrhythmia prediction. Overall, our contribution substantially advances the current methodology for predicting arrhythmia and caters to clinical practitioners' needs by providing an accurate and fully transparent tool to support arrhythmia prediction decisions.