Enabling Point of Care Diagnosis of Epilepsy in Underserved Populations

Epilepsy is a major source of death and disability worldwide. Specialists diagnose epilepsy by integrating the clinical history with specialized tests, particularly the electroencephalogram (EEG). Accurate diagnosis is critical because effective treatments prevent recurrent seizures in “positive” cases, and in “negative” cases alternative (e.g. cardiac) interventions may be necessary. Nevertheless, most epilepsy diagnoses in the U.S. are made by non-specialists, and worldwide, most people with epilepsy lack access to diagnostic and treatment resources. There exists a critical need to increase access to reliable epilepsy diagnostic testing. Our group previously showed that a deep learning (DL) AI algorithm was able to identify epileptiform discharges (ED) in the EEG, the key physiologic biomarker of epilepsy, more accurately than individual epilepsy specialists. However, before such an AI can be used by non-specialists to accurately diagnose patients with suspected epilepsy, further work remains. Our central hypothesis is that point-of-care EEG technology, automated interpretation of EEG, and a simple questionnaire can enable non-experts to deliver accurate diagnostic services to patients with suspected epilepsy. The work for this project will be accomplished through three specific aims: SA1) To render interpretation of routine EEG accessible in underserved settings, we will develop a comprehensive AI approach to detecting epileptiform abnormalities. We will also train the new AI to classify the type of epilepsy based on the EEG. This aim will require obtaining annotations of 10K EEGs. These will also be made available as a resource to the research community. SA2) To allow non-experts to diagnose epilepsy as accurately as experts, we will develop a risk score that uses clinical history and AI-EEG to predict seizure recurrence. The score will be developed from prospectively collected data from 1000 patients. SA3) Aim 3: To validate and optimize our system for diagnosing epilepsy, we will test our point-of-care method in the target underserved patient groups in both a Boston-based Emergency Department as well as partner with health care workers in Guinea, an underserved area, to field-test the system in 600 new participants. This work will provide four key deliverables. 1) Ability to obtain EEGs and automated expert-level EEG interpretations in real-time at the point of care; 2) a unique, massive, public annotated dataset; 3) an epilepsy risk questionnaire that non-experts can administer and combine with automated EEG interpretation to make an accurate clinical diagnosis; 4) proof-of-concept for providing epilepsy diagnostic services in underserved areas. Project Number: 1R01NS144052-01 | Fiscal Year: 2026 | NIH Institute/Center: National Institute of Neurological Disorders and Stroke (NINDS) | Principal Investigator: Farrah Mateen | Institution: NORTHWESTERN UNIVERSITY, CHICAGO, IL | Award Amount: $600,646 | Activity Code: R01 | Study Section: Clinical Informatics and Digital Health Study Section[CIDH] View on NIH RePORTER: https://reporter.nih.gov/project-details/11204622