Video-based automatic seizure detection in pharmacoresistant epilepsy: A prospective exploratory study

Area of Science:

• Neurology
• Medical Technology
• Artificial Intelligence

Background:

• Pharmacoresistant epilepsy presents diagnostic challenges, often requiring prolonged video-electroencephalography (EEG) monitoring.
• Nocturnal motor seizures can be difficult to detect and classify accurately.
• Automated seizure detection devices offer a potential solution for continuous monitoring and objective assessment.

Purpose of the Study:

• To evaluate the diagnostic yield and clinical utility of an AI video-based seizure detection device (SDD) in patients with pharmacoresistant epilepsy.
• To assess the accuracy of the SDD in classifying nocturnal motor events as epileptic or non-epileptic.
• To compare the device's performance for major focal motor seizures versus subtle focal motor seizures.

Main Methods:

• Prospective recruitment of patients with focal epilepsy and pharmacoresistance undergoing inpatient video-EEG monitoring.
• Home-based nocturnal monitoring using the SDD for a median of 15.5 nights.
• Analysis of captured video recordings by clinical experts and assessment of clinical utility based on pre-specified measures.

Main Results:

• The SDD demonstrated a diagnostic yield of 55.0% and clinical utility in 40.0% of registration sessions.
• AI classification showed no significant difference compared to clinical experts' consensus.
• Diagnostic yield and clinical utility were significantly higher for major focal motor seizures (81.8% and 63.6%) than for subtle focal motor seizures.

Conclusions:

• The AI-powered SDD is a useful tool for evaluating patients with pharmacoresistant epilepsy, especially those with major focal motor seizures.
• The device can facilitate the diagnostic process and potentially guide anti-seizure treatment adjustments.
• The SDD accurately classifies major focal motor seizures, aiding in the differentiation of epileptic and non-epileptic events.