Design of a Smart Alert System for Epileptic Seizure Detection

Abstract

This work addresses the problem of detecting seizures experienced by a epileptic patients. Earlier solutions to this problem are mostly based on detectingthe jerking motion due to seizures which are less accurate because of large number of false detection alarms. We seek a alternative that can sense epileptic seizures by sensing the variations in skin conductance of patient due to variations in electrical activity of brain. Since this moniters the electrodermal activity it provides high detection accuracy. The Galvanic Skin Response(GSR) sensor is used to detect this changes in the skin conductance. The Hybrid Neural Network which is a combination of Convolutional Neural Network(CNN) and Recurring Neural Network(RNN)is used fordataset classification of the GSR sensor output into occurrence of epileptic seizures and ordinary motions. Thus GSR sensors could potentially be used to detect epileptic seizures as the successfully detected the simulated seizures with small errors only in differentiating ordinary motion and no motion and give emergency alert using wifi module ESP8266 in nodeMCU.
Keywords:Epilepsy, Seizures, Galvanic Skin Response, NodeMCU, Hybrid Neural Network