Abstracts

Rationale: There is an unmet need for reliable, easy to use detection devices to alert for clinically urgent seizures. Methods: The NightWatch bracelet (Fig. 1) combines photoplethysmography (PPG) and accelerometry (ACM) to assess rapid changes in heart rate and/or excessive motion patterns to detect the full spectrum of nocturnal clinically urgent seizures: tonic-clonic, hypermotor and prolonged generalised tonic seizures (>30 seconds). The bracelet is worn at the upper arm and wirelessly sends alarms to a central station using DECT technology. At the time of submission of this text, the performance of the system was evaluated in 10 patients with refractory epilepsy and learning disabilities admitted to residential care department of the participating Dutch epilepsy centres. Nocturnal video and/or audio monitoring was used to review all alarms generated by caregivers or the device and to screen 10% of all nights irrespective whether an alarm was generated or not. A team of specialised epilepsy nurses and neurologists blinded to the type of the alarm, reviewed all alarms to determine the epileptic nature of the event and to validate the seizure classification. Results: We present the interim results of the performance of the system (Table 1); the algorithms will be detailed elsewhere. Mean age of the ten patients was 38 (range 22-63 years) were assessed for a variable period of 35-86 nights/patient; one of the 10 patients had no nocturnal clinically urgent seizures during this period. The Nightwatch bracelet was well tolerated. The overall sensitivity of the Nightwatch system for detecting severe seizures was 94% (range 67%-100%). The overall positive predictive accuracy (true alarm rate) was 62 % (range 39%- 100%); the single subject without seizures had .5 false alarm/night (28 in 52 nights). Conclusions: Multimodal algorithms combining PPG and ACM is a promising way to detect clinically urgent seizures. The false alarm rate is however variable and may require a tailored algorithm in a subset of subjects. Funding: This study is made possible by a grant from the dutch NWO-ZONMW TOP (300040003); the NutsOhra fund grant 1203-050 and a donation by the dutch epilepsy fund. The sensor hardware was developed and made available by the Livassured company, Vught, the Netherlands.