19407
Overview
Epilepsy affects ~50 million people worldwide, with 30% suffering from drug-resistant epilepsy (DRE) that drives persistent seizures, high mortality, and frequent misdiagnosis. Current EEG-based diagnostics are expert-dependent, slow, and inconsistent. This technology applies unsupervised machine learning to automatically detect interictal epileptiform discharges (IEDs) and patient-specific network markers highly correlated with clinical outcomes. Automated, AI-driven EEG analysis enables accurate, efficient, and scalable diagnostic workflows.
Applications
- Improved accuracy and consistency of epilepsy diagnosis
- Personalized network-level biomarker-guided monitoring of treatment response and disease progression
- Support for pre-surgical evaluation in drug-resistant epilepsy
- Scalable large-scale EEG analysis for clinical and research settings for novel insights for clinicians
- Potential future real-time prediction of incoming seizure using the personalized network level biomarkers, allowing preventive actions
Differentiation
- Unsupervised AI: Reduces expert bias & manual workload while improving accuracy
- Personalized markers: Provides individualized measures linked to clinical outcomes
- Scalable: Enables efficient, high-throughput EEG analysis
Development Stage
- A prototype ML algorithm was implemented and benchmarked against neurologist annotations, achieving high recall and detecting significantly more events than experts.
- Initial correlations with clinical markers were demonstrated.
- Next steps include extensive validation across larger patient datasets and development of full epileptiform network detection models in collaboration with Rabin Medical Center.
Contact for more information
Dr. Vered Pardo Yissar
Senior Director of Business Development, Exact Sciences