Development of a Wearable 3-Risk Factor Accumulated Epileptic Seizure Detection System with IoT Based Warning Alarm

In this work, a fully-functional prototype of a wristworn device has been developed for detecting convulsive epileptic seizure. A 3-risk factor based detection scheme employing accelerometer derived muscle spasm, PPG acquired heart rate variability and body temperature variation has been proposed by utilizing the distinctive change in the patterns of these extracerebral modalities during seizure. Since the reproduction of these specific combination of patterns is less likely to occur in case of nonseizure events during patients' regular activities, the developed device identifies the seizure occurrence quite efficiently. An advanced warning alarm system is also incorporated in the design via IoT intervention. A master Bluetooth module is used for transmitting data and another Bluetooth module is used as a slave to notify the nearby people regarding the threat of the patient as well as drive a loud alarm. The user-friendliness is also ensured based on the feedback from volunteers. With a sensitivity of 85%, a missed alarm rate of 15% and a false alarm rate of 26.09%, the device has definitely demonstrated promising performance for the detection of convulsive epileptic seizure. Still, there are scopes of further improvement in terms of efficacy by algorithm optimization and deep learning involvement.