Automated seizure detection using EKG.

Area of Science:

• Neurology
• Biomedical Engineering
• Medical Devices

Background:

• Epileptic seizures often cause heart rate changes.
• Current seizure detection methods like electroencephalogram (EEG) and electrocorticogram (ECoG) can be invasive or cumbersome.
• Automated seizure detection aims to improve patient monitoring and management.

Purpose of the Study:

• To evaluate the feasibility and accuracy of an electrocardiographic (EKG)-based algorithm for automated epileptic seizure detection.
• To compare the performance of EKG-based detection with electrocorticogram (ECoG) monitoring.
• To assess the potential clinical utility of EKG for seizure detection and as an electronic seizure diary.

Main Methods:

• Analyzed 241 clinical seizures from 81 subjects undergoing invasive monitoring.
• Developed and tested an EKG-based seizure detection algorithm with varying sensitivity settings (threshold T, duration D).
• Compared EKG algorithm performance against a validated ECoG algorithm, analyzing false negatives, false positives, and timing of seizure onset detection.

Main Results:

• The EKG algorithm detected 98% of seizures at its most sensitive settings (T: 1.15, D: 0s).
• False negative rates increased to 14% at higher settings (T: 1.3, D: 5s).
• Potential false positive detections were significantly reduced with higher settings (1.1/h vs 9.5/h), with a substantial portion linked to seizures or epileptiform discharges.

Conclusions:

• Automated EKG-based seizure detection is a feasible and potentially valuable clinical tool.
• EKG offers advantages in ease of acquisition, processing, and signal quality compared to EEG/ECoG.
• EKG monitoring can serve as a cost-effective "electronic" seizure diary, improving accuracy over patient-reported logs.