Abstracts

Rationale: Automated seizure detection may prevent complications, especially during the night. Most current systems focus on the detection of tonic clonic seizures using movement sensors, but no long term data are yet available. Methods: In this long-term, multi-centre, prospective trial we validated a novel device combining heart rate (photoplethysmography) and movement (3-D accelerometry) algorithms in people with epilepsy and intellectual disability. Participants with > 1 major motor seizure per month wore a bracelet with both sensors (Nightwatch ®) at the upper arm for 2-3 months during the night. Video recordings of all reported events (as indicated by device-generated alarms or nurse diaries) and random screening of 10% of all nights were reviewed and classified as 'major motor seizures' (requiring an alarm), minor seizures (no alarm) or no seizures (no alarm). Major motor seizure included: tonic-clonic seizures, generalized tonic >30 s seizures, hyperkinetic seizures or clusters (> 30 min.) of myoclonus or short tonic episodes. Reliability was tested by interobserver agreement (Fleiss’ kappa). The performance of the Nightwatch was assessed (sensitivity; positive predictive value and false negative detection rate) and, if possible, compared to a simultaneously used bed sensor (Emfit®). Results: Twenty-eight participants (1826 nights, 809 major motor seizures) completed the study. Interobserver agreement (presence vs. absence of 'major motor seizure') was 0·77. Nightwatch (NW) yielded a high median sensitivity of 86% (95% CI 76 – 96), low false negative detection rate (0·04 per night, 95%, CI 0·00 - 0·08) and moderate positive predictive value (49%, 95% CI 37 – 62). In those 14 subjects that used both NW and a bed sensor, Nightwatch showed a better sensitivity (median difference, 58·1%, p < 0·001), lower false negative detection rate (-0.21 per night, p < 0·05) and comparable positive predictive value (2·3%, n.s.) Conclusions: Algorithms combining heart rate and movement improved nocturnal seizure detection in a residential population with severe refractory epilepsy. Funding: Netherlands Organisation for Scientific Research (ZonMW), Nuts-Ohra Foundation, Dutch Epilepsy Fund.