Internet of Things for Epilepsy Detection in Patients

Abstract

This article details the design of a web application working alongside a device that uses sensors to capture and transmit data related to the brain’s activity (normal or abnormal) from a patient who experiences symptoms of epilepsy. The purpose is to prevent this disease from causing harm and irreversible effects on that specific population. The sensors are powered by a mobile device that connects via Bluetooth when changes are detected. Then, a signal is transmitted which is analyzed using neural networks for the debugging and processing of the information. A decision is then made regarding the state of the patient who could be suffering from an epileptic seizure. In such case, a report is issued in order to save his life. Specific characteristics found in people with critical episodes of epilepsy are combined with a hybrid system consisting of a logical controller based on an Adaptive System of Neural-Diffuse Inference (ANFIS). This study concludes that the model validated with a database including 198 signs in the years 2010, 2011 and 2012 has an accuracy of 95.5% in diagnosing or predicting an epileptic seizure. The performance matches the accuracy found in other techniques.