Identifying heart arrhythmias through multi-level algorithmic processing of ECG on edge devices

Accurate and timely detection of arrhythmia is crucial to reduce treatment times and ultimately prevent serious life-threatening complications such as the incidence of a stroke, especially during the diagnostic process in settings with limited access to cardiologists, such as hospital emergency departments. The proposed lightweight solution uses a novel classifier consistently designed and implemented based on a 2D convolutional neural network (CNN) and properly optimized in terms of storage and computational complexity thus, making it suitable to be deployed on edge devices capable of operating in the hospital emergency department providing privacy and portability. Experiments on the MIT-BIH arrhythmia database, show that the proposed 2D-CNN obtains overall accuracy of 95.2%, mean sensitivity of 95.2%, specificity of 98.8%, and One-vs-Rest ROC-AUC score of 0.9897. Moreover, results based on the Jetson Nano metrics show that the proposed method achieved an excellent performance and would be particularly useful in the clinical practice for continuous real time (RT) monitoring scenarios.