Mayo Clinic AI tools accelerate seizure hot spot detection to shorten drug-resistant epilepsy monitoring

Drug-resistant epilepsy patients must undergo brain electrode implantation surgery followed by weeks of monitoring to localize seizure-causing tissue. High-frequency brain waves are difficult to detect due to short duration, low amplitude, and environmental noise. The existing manual data-cleaning process is labor-intensive and slows the path to curative surgery.

Mayo Clinic researchers developed AI tools to more rapidly and accurately pinpoint seizure hot spots by automating detection of high-frequency brain waves and removal of corrupted data points. Future work aims to transform the framework into a fully digital, real-time intraoperative system that provides feedback on epileptic tissue location during electrode implantation surgery itself.

The AI tools enable faster, more accurate identification of seizure-causing brain tissue, accelerating eligibility for targeted tissue removal surgery critical for achieving seizure freedom. Faster identification reduces the duration of post-implantation monitoring stays, lowering infection risk — which is five times higher in children than adults during prolonged epilepsy monitoring unit stays.

• Automating intracranial EEG data cleaning with AI directly shortens the time patients spend in high-risk monitoring units.
• Reducing pediatric monitoring duration has outsized safety benefits given their 5x elevated infection risk.
• Real-time intraoperative AI feedback during electrode surgery represents the next clinical frontier for epilepsy localization.