Multimodal nocturnal seizure detection: Do we need to adapt algorithms for children?

Area of Science:

• Neurology
• Biomedical Engineering
• Pediatric Epilepsy

Background:

• Multimodal seizure detection devices, initially validated in adults, require assessment in pediatric populations.
• Epilepsy in children, particularly those with intellectual disabilities, presents unique monitoring challenges.
• Nocturnal seizure detection relies on physiological signals like heart rate and movement.

Purpose of the Study:

• To evaluate the performance of a multimodal seizure detection device in a pediatric cohort.
• To assess the device's sensitivity and positive predictive value (PPV) in children with epilepsy.
• To investigate methods for reducing false alarms in pediatric seizure detection.

Main Methods:

• A multicenter, prospective, video-controlled cohort study involving children with epilepsy and intellectual disability.
• Participants wore a Nightwatch bracelet for nocturnal seizure detection over three months.
• Seizure classification involved reviewing video recordings of alarms and nurse diaries.

Main Results:

• Initial testing showed a sensitivity of 79.9% and a PPV of 26.7%, with a high false alarm rate (0.2/h), primarily due to heart rate detection in awake children.
• An adaptation to the algorithm, incorporating body position (horizontal), significantly improved performance.
• Post-adaptation, the device achieved a mean PPV of 55.5% and a reduced false alarm rate of 0.08/h, while maintaining a sensitivity of 79.4%.

Conclusions:

• Seizure detection devices using heart rate and movement demonstrate comparable sensitivity in children and adults.
• Children exhibit a higher false alarm rate, often linked to being awake and body movement.
• Adjusting for body position effectively reduces false alarms in pediatric seizure detection, approaching adult levels.