An Edge-computing Platform for Low-Latency and Low-power Wearable Medical Devices for Epilepsy

Abstract

Epilepsy is a neurological disorder that affects 1% of people globally. The development of a portable, low-power, and low-latency wearable sensor is a growing need to address epilepsy. An edge-computing-based wearable sensor has been presented that uses a pulse exclusion mechanism (PEM) and a random forest classifier to identify seizures at a reduced delay and minimal power consumption. Datasets recorded from the scalp electrode are utilized to demonstrate the feasibility of using the method as a wearable medical device. Including the edge-IoT platform in place of cloud IoT offers a considerable reduction in system latency. The optimized edge-computing platform reduces power usage significantly compared to existing methods. The reduced latency and battery usage make the proposed device faster and more energy-efficient, which may be useful for low-power wearable devices.