Automated seizure detection using wearable devices: updated systematic review and meta-analysis of tonic-clonic and...

Wearable devices have emerged as promising tools for continuous seizure monitoring in patients with epilepsy. This study aimed to update the evidence on the diagnostic performance of non-invasive wearable sensors for seizure detection, extending the previous systematic review by Naganur et al., which included studies published up to November 10, 2021.

A systematic review and meta-analysis were conducted using studies published up to July 15, 2025. Eligible studies used video-electroencephalographic (video-EEG) monitoring as the reference standard and reported sensitivity and false alarm rate (FAR) for wearable seizure detection devices. Included devices comprised wrist-worn and surface-worn sensors measuring parameters such as accelerometry, heart rate, and electrodermal activity.

Thirty-one studies evaluating tonic-clonic seizures (TCS) were included, comprising 2,128 patients and 112,542 h of video-EEG monitoring. The pooled sensitivity for TCS detection was 89.9% (95% CI: 85.0-93.9%), with a FAR of 1.43 per 24 h (95% CI: 1.22-1.64). Wrist-worn devices demonstrated slightly higher sensitivity (90.3%) but higher FAR (1.7/24 h) than surface devices (88.1% sensitivity; 0.8/24 h FAR). For focal seizures, nine studies involving 660 patients and 15,056 monitoring hours were included. The pooled sensitivity was 73.5% (95% CI: 57.4-87.0%) and the FAR was 2.85 per 24 h (95% CI: 1.40-5.81). Excluding one study that reported only overall sensitivity increased focal seizure sensitivity to 78.4%.

Wearable devices demonstrate high sensitivity for detecting tonic-clonic seizures, although false alarms remain a challenge. Detection performance for focal seizures is substantially lower, highlighting the need for improved algorithms and multimodal sensing approaches. These findings support the continued development of wearable technologies for real-world epilepsy monitoring and management.