Seizure detection based on heart rate variability using a wearable electrocardiography device

Area of Science:

• Neurology
• Biomedical Engineering
• Cardiology

Background:

• Current nonconvulsive seizure detection methods have limitations in sensitivity and false alarm rates.
• Wearable electrocardiography (ECG) devices offer a noninvasive approach to monitoring physiological changes during seizures.
• Heart rate variability (HRV) analysis is a potential biomarker for detecting autonomic nervous system disturbances associated with seizures.

Purpose of the Study:

• To evaluate the feasibility and accuracy of seizure detection using HRV analysis from a wearable ECG device.
• To assess the performance of automated HRV algorithms in identifying both convulsive and nonconvulsive seizures.
• To compare algorithm performance against expert-marked seizure time-points from long-term video-EEG monitoring.

Main Methods:

• A prospective phase 2 study involving 100 patients undergoing long-term video-EEG monitoring.
• Continuous ECG recording using a dedicated wearable device.
• Off-line computation and analysis of 26 automated HRV algorithms for seizure detection, blinded to other data.

Main Results:

• The best HRV algorithm detected 53.5% of patients with seizures as responders.
• Detection sensitivity was 93.1% for all seizures and 90.5% for nonconvulsive seizures among responders.
• Low false alarm rate (1.0/24h) and median detection latency of 30 seconds were observed.

Conclusions:

• Automated HRV analysis using wearable ECG devices demonstrates high sensitivity for detecting seizures, including nonconvulsive types.
• The approach is feasible and accurate, particularly in patients exhibiting significant ictal autonomic changes.
• This technology offers a promising noninvasive tool for continuous seizure monitoring.