Abstracts

Rationale: Seizure apps using biosensors and mobile devices may be used to develop non-EEG seizure detection and help support patients with epilepsy. We implemented a national study called EpiWatch, which uses the Apple Watch and encrypted data transmission to track heart rate, movements, and responsiveness during seizures, with the goal of using this data to develop seizure detection and to provide patients with an epilepsy support system. Methods: Participants were electronically consented using the ResearchKit platform; they then received emails with PDFs of signed consents. The participants’ seizure data and health survey data were encrypted, transmitted, and de-identified for research use via a data backend server system. Participants activated seizure tracking during auras or had caregivers activate tracking. The app collected seizure sensor data for 10 minutes, with participants tested for responsiveness each minute. Participants labeled their seizure data in a brief post-seizure survey. Family or caregivers could receive SMS alerts during seizures. We performed preliminary analysis of participant demographics, seizure types and changes in heart rate, movements and alertness during seizures. Results: 501 participants enrolled in the EpiWatch study during an initial 9-month period. Participants were consented from nearly all states with a mean age of 32 (range of 16-73) and a broad range of ethnicities. 45% of participants tracked seizures (mean 6.9 seizures tracked per participant) with a total of 1557 seizures recorded.  Participants’ seizure types were typical of those expected for a national epilepsy population: 34% complex partial seizures, 12% tonic clonic seizures, 18% auras, 17% simple partial seizures, 13% were labeled as absences; the remaining had tonic, atonic, and myoclonic seizures. During tonic-clonic seizures, heart rates increased >30% in 38% and >50% in 23% of participants’ seizures. Heart rate changes were more variable with complex partial seizures, with >15% increases in heart rates during 42% and decreases in 14% of seizures. Conclusions: Our study demonstrates that it is possible to rapidly implement a national study which collects biosensor and response data during seizures using a mobile device. Participants are representative of a typical national cross-section of patients with epilepsy (seizure types, ages, ethnicities, locations). Watch biosensors captured heart rate accelerations, particularly during tonic-clonic seizures. This data, along with accelerometer and responsiveness data are being used to test seizure detection algorithms. The EpiWatch research app also provides patient support. Funding: InHealth Research Grant, Johns Hopkins University