Location

MIT Media Lab, E14-633

Description

Epilepsy is a neurological disorder that predisposes individuals to have recurrent, unprovoked seizures. The apprehension about injury, or even death, resulting from a seizure often overshadows the lives of those unable to achieve complete seizure control. Moreover, people with uncontrolled seizures are at greatest risk for sudden unexpected death in epilepsy (SUDEP). 

This thesis describes the development of a wrist-worn biosensor and investigates its clinical utility in the assessment of epileptic seizures. The sensor provides unobtrusive and continuous measurements of electrodermal activity (EDA), a sensitive index of sympathetic autonomic activity, and accelerometry (ACM) over extended periods of time. 

Using the wrist-worn biosensor and time-frequency mapping of heart rate variability, we analyzed the autonomic alterations that accompany epileptic seizures. We observed a critical window of severe autonomic dysregulation after generalized tonic-clonic seizures that may be relevant in the pathogenesis of SUDEP. Importantly, the degree of autonomic disturbance was strongly correlated with the duration of post-ictal generalized electroencephalographic (EEG) suppression, an objective marker for the risk of SUDEP. 

Lastly, this thesis presents a novel algorithm for generalized tonic-clonic seizure detection using EDA and ACM. The algorithm was tested on 4213 hours (176 days) of continuous recordings from 80 patients containing a wide range of ordinary daily activities and detected 15/16 (94%) tonic-clonic seizures with an average false alarm rate of ≤ 1 per 24-hour period. We anticipate that the proposed wrist-worn biosensor will provide an ambulatory seizure alarm and improve the quality of life of patients with uncontrolled tonic-clonic seizures.

Participant(s)/Committee

John V. Guttag, PhD., Richard J. Cohen, MD, PhD.,  Tobias Loddenkemper, MD