Detection of Bilateral Tonic-Clonic Seizures Using Miniaturized Wearable Electromyography-Accelerometry Sensors

Area of Science:

• Biomedical Engineering
• Neurology
• Wearable Technology

Background:

• Epilepsy monitoring often relies on subjective observations or cumbersome equipment.
• Accurate and timely seizure detection is crucial for patient safety and intervention.
• Wearable sensors offer a promising avenue for continuous, unobtrusive physiological signal monitoring.

Purpose of the Study:

• To develop and evaluate a seizure detection algorithm using surface electromyography (sEMG) and accelerometry (ACC) signals.
• To assess the algorithm's performance with varying numbers of miniaturized wearable sensors.
• To validate the algorithm's efficacy in detecting generalized tonic-clonic seizures (GTCS) and focal-to-bilateral tonic-clonic seizures (FBTCS).

Main Methods:

• Continuous sEMG-ACC signals were collected from patients using eight bilateral wearable sensors.
• An extreme gradient boosting classifier was trained to identify seizure epochs.
• Performance was evaluated using patient-wise nested cross-validation and assessed on a non-seizure cohort for specificity.

Main Results:

• The algorithm achieved 100% sensitivity for detecting GTCS and FBTCS with a dual-sensor setup (right biceps brachii and left tibialis anterior).
• The dual-sensor setup demonstrated a low false alarm rate (0.12 FAR/24h) and a median detection latency of 22 seconds.
• Specificity was high, with a false alarm rate of 0.06 FAR/24h on data from patients without seizures.

Conclusions:

• The developed algorithm effectively detects GTCS and FBTCS in an epilepsy monitoring unit, even with a reduced sensor count.
• This wearable sensor-based seizure detection system has the potential to enable timely interventions in outpatient settings.
• The technology could enhance safety and independence for individuals with epilepsy.