Automated Detection of Convulsive Seizures Using a Wearable Accelerometer Device.

Area of Science:

• Biomedical Engineering
• Neurology
• Wearable Technology

Background:

• Epileptic seizure detection traditionally relies on resource-intensive video/electroencephalography (EEG) monitoring in clinical settings.
• Expert analysis of video/EEG data is labor-intensive and limits continuous, remote patient monitoring.
• There is a need for accessible, non-invasive methods for ambulatory seizure detection.

Purpose of the Study:

• To develop and evaluate a wireless remote monitoring system for epileptic seizure detection using a single wrist-worn accelerometer.
• To assess the system's sensitivity to various convulsive seizure types and its capability for detecting short-duration seizures.
• To establish a non-invasive, continuous monitoring solution for ambulatory seizure detection.

Main Methods:

• Extraction of time-domain and novel Poincar´e plot-based features from accelerometer data during active movement events.
• Feature selection using Receiver Operating Characteristic (ROC) curve analysis to identify the most discriminative features.
• Classification of non-seizure and seizure events using Kernelized Support Vector Data Description (SVDD).

Main Results:

• The system evaluated on 5,576 hours of data from 79 patients detected 86.95% (40/46) of convulsive seizures (GTCS, PNES, CPS) in 20 patients.
• The algorithm achieved a comparable performance for generalized tonic-clonic seizure (GTCS) detection, with 95.23% sensitivity and a 0.64 false alarm rate per 24 hours.
• A total of 270 false alarms were recorded over the evaluation period (1:16 false alarms/24h).

Conclusions:

• The proposed wrist-worn accelerometer-based system demonstrates significant potential for ambulatory monitoring of convulsive epileptic seizures.
• This wearable technology offers a timely, non-invasive, and less resource-intensive approach to continuous seizure assessment.
• The findings support the development of wearable devices for improving the management and understanding of epilepsy outside clinical settings.