Foundation Models for Epileptogenic Zone Identification in Drug-Resistant Epilepsy

2026-06-21

AI Summary

This study addresses the suboptimal seizure control rate—below 50%—in patients with drug-resistant epilepsy, often due to inaccurate localization of the epileptogenic zone (EZ) prior to surgery. The authors propose EpiiSLM, a dual foundation model system that, for the first time, leverages interictal sleep stereoelectroencephalography (sEEG) signals to anchor EZ biomarkers, enabling precise localization using only a single night of seizure-free data and substantially reducing invasive monitoring duration. By integrating a signal foundation model with a language foundation model, EpiiSLM synthesizes multimodal clinical information to deliver interpretable EZ identification. Evaluated via leave-one-patient-out cross-validation on 104,990 minutes of sEEG data, the method achieved an internal contact-level positive predictive value (PPV) of 0.978 and 100% region-level accuracy; in external testing, it attained a PPV of 0.857, representing a statistically significant 15.1% improvement over conventional seizure onset zone (SOZ)-based approaches (p<0.05).

Abstract

Accurate identification of the epileptogenic zone (EZ) is essential for seizure freedom after resective surgery in drug-resistant epilepsy, yet seizure freedom rates remain below 50%. We developed EpiiSLM, a dual foundation model system for EZ identification with stereo-electroencephalography (sEEG), by training a signal foundation model on 104,990 minutes of sEEG recordings from the Montreal Neurological Institute & Hospital, while leveraging all recordings regardless of surgical outcome and anchoring EZ biomarker extraction on non-epileptic signals. A language foundation model then integrates sEEG-derived outputs with multimodal clinical information to produce interpretable predictions. Under leave-one-patient-out evaluation, EpiiSLM achieved 0.978 contact-level positive predictive value (PPV), outperforming the seizure onset zone(SOZ)-as-EZ baseline by 15.1% (p < 0.05), and 100% region-level accuracy; on an external dataset, EpiiSLM achieved 0.857 contact-level PPV. EpiiSLM requires only one night of interictal sleep data, suggesting potential to reduce invasive sEEG monitoring duration and improve surgical outcomes.

Problem

• epileptogenic zone
• drug-resistant epilepsy
• sEEG
• seizure freedom
• resective surgery

Innovation

• foundation model
• epileptogenic zone
• stereo-EEG
• multimodal integration
• interpretable AI