A Deep Learning Model for Arrhythmia Detection Using Ultra-Edge IoT Sensors

The proliferation of Internet of Things (IoT) devices in edge networks has highlighted their limitations in computational resources for ultra-edge analytics. This paper explores the advancement beyond traditional edge analytics, focused primarily on user-smartphones, by embedding intelligence into ultra-edge IoT sensors. The study particularly investigates smart IoT sensors for arrhythmia detection using Electrocardiogram (ECG) traces, exemplifying an enhanced intelligence application in mobile health. This paper introduces the E2DC-Net, a novel deep learning model leveraging a Multi-Scale Convolutional Neural Network (MSCNN), designed to enhance arrhythmia detection in ultra-edge IoT sensor networks. The methodology focuses on the MSCNN's ability to process complex ECG data by capturing multi-scale features, thereby improving diagnostic accuracy and reliability.