Feasibility Study of Seizure Detection using Commercial Smartwatches in EMU

This study explores the feasibility of using commercial smartwatches and smartphones in the Epilepsy Monitoring Unit (EMU) for seizure data collection and development of Machine Learning models for seizure detection. The accelerometer and heart rate sensors in a Garmin Venu 3 smartwatch are used for data recording and a commercial smartphone is used as a wireless Hub for user registration, data storage and management. We addressed several challenges in developing a remote data collection pipeline, which includes raw sensor data synchronization, CSV file generation, data upload, extended battery life, as well as uninterrupted operation through foreground services and programmed background processes. With an optimized mobile application, continuous data collection, storage and transfer have been demonstrated for more than 5 days.