Abstracts

Rationale: Monitoring for tonic-clonic seizures (TCS) is important for enhancing safety, promoting independence, and avoiding sudden unexpected deaths in epilepsy (SUDEP). Any system for TCS monitoring should be highly sensitive, present a low false alarm rate (FAR) and provide alerts to caregivers with a low latency across use in both inpatient and ambulatory environments. Ideally, these devices should also be non-invasive, multifunctional and avoid stigma. This preliminary study describes the performance characteristics of a custom-built seizure monitoring application implemented on a consumer wearable (Apple Watch), tested in both inpatient and ambulatory environments.

Methods: Data was initially collected from 340 patients in 4 Epilepsy Monitoring Units (EMUs), and 21 ambulatory users (13 outpatients with epilepsy, 8 normal controls without epilepsy). Accelerometer and heart rate signals were recorded with a custom application developed for Apple Watch. Seizures in the EMUs were validated with video-electroencephalography (vEEG), while ambulatory user seizure events were self-reported and not included in the training set as they were difficult to validate. This yielded a dataset of 20,388 hours including 79 TCS (58 EMU patients), and 5,642 seizure-free hours from ambulatory users (outpatients and normal controls). This data was used to train a novel classifier in an offline environment that was subsequently implemented on Apple Watch, as part of a custom-built seizure monitoring application. Prospective testing was performed on 85 unique EMU patients and 15 ambulatory users (9 with outpatients, 6 normal controls). EMU patients were blinded to seizure detections, and seizures were validated with vEEG. Ambulatory users were unblinded to seizure detections, and seizures were self-reported or retrospectively identified through independent bio-signal analysis. The testing dataset was 4,279 hours in the EMU with 19 seizures (15 patients) and 6,735 hours in outpatients with 10 self-reported seizures (3 patients). 

Results: The prospective testing resulted in a positive percent agreement (PPA) of 100%, an FAR of 0.05 per day in the EMU (positive predictive value, PPV, of 68%) and 0.13 per day in ambulatory users (PPV of 22%). A single outpatient was responsible for 8 of 31 total false alarms. The FAR for all other ambulatory users excluding this outpatient was 0.10 per day. Mean detection latency was 37.38 s (stdev = 13.24s) n the EMU and 32.07 s (stdev = 10.22s) in ambulatory users. 

Conclusions: Our Apple Watch based TCS monitoring application had high sensitivity while maintaining false alarm rates substantially lower than previously reported in both inpatient and ambulatory environments. To our knowledge, this is the first prospective multi-center study of TCS monitoring using a consumer wearable device. We believe using a seizure monitor on a consumer wearable has the potential to promote safety in PWE, as these multifunctional devices are already well adopted and familiar to users, may be used independently of any paired smartphone, and avoid the potential stigma associated with bespoke monitoring devices.

Funding: Please list any funding that was received in support of this abstract.: Band Foundation and JHU Department of Neurology.