ECT seizure duration: reliability of manual and computer-automated determinations

Area of Science:

• Neuroscience
• Clinical Medicine

Background:

• Monitoring electroencephalographic (EEG) and electromyographic electroconvulsive therapy (ECT) seizure duration is crucial in clinical practice.
• Accurate automated seizure duration determination is highly valuable for routine ECT assessments.

Purpose of the Study:

• To assess the reliability of computer-automated seizure duration determinations from the Thymatron Model DGx ECT machine.
• To explore factors influencing the reliability of automated and experienced human raters' seizure duration assessments.

Main Methods:

• Comparison of seizure duration ratings between the Thymatron DGx automated system and experienced human raters.
• Analysis of factors affecting rating reliability, including artifact, postictal suppression, and EEG seizure endpoint characteristics.

Main Results:

• Experienced human raters demonstrated very high interrater reliability, significantly exceeding that of the automated system.
• Reliability for both automated and human ratings decreased with artifact, poor postictal suppression, or gradual EEG seizure endpoints.
• Automated system reliability was particularly sensitive to these factors, varying from 0.68 to 0.999.

Conclusions:

• Experienced human raters are more reliable than the Thymatron DGx automated system for ECT seizure duration monitoring.
• Clinical factors significantly impact the reliability of both automated and human seizure duration assessments.
• Further refinement of automated systems is needed to improve accuracy in complex clinical scenarios.