AI Model Outperformed Neurologists in Epilepsy Detection

Researchers developed an AI tool that improved clinical diagnostic accuracy for epilepsy by over 14 percent.

Updated on Oct. 8, 2026 in Artificial Intelligence

Isometric editorial illustration of a stylized fiber-optic neural model and a medical sensor head, representing advanced epilepsy diagnostic technology. — Researchers developed an AI tool called EpiScreen that improved clinical diagnostic accuracy for epilepsy by over 14 percent in recent trials. AI Illustration. Upload story photo >

Live Poll

Would you trust an AI tool to assist your doctor in identifying a potential health condition?

The new AI model, named EpiScreen, has demonstrated superior diagnostic capabilities compared to human experts. By analyzing electronic health records, the tool helped neurologists identify epilepsy cases more accurately.

Why it matters

Epilepsy is frequently misdiagnosed, leading to significant diagnostic delays for patients. This technology offers a cost-effective alternative to traditional, resource-intensive video-electroencephalography diagnostics.

EpiScreen achieved an area under the curve (AUC) score of 0.875 on the MIMIC-IV dataset and a higher score of 0.980 on a cohort from the University of Minnesota.

The players

EpiScreen

This is a new artificial intelligence model designed to assist clinicians in the detection of epilepsy using patient electronic health records.

University of Minnesota

This public research university provided the patient cohort data used to test the diagnostic accuracy of the new AI model.

The details

Researchers fine-tuned large language models on labeled clinical notes to create the diagnostic tool. In clinical trials, neurologists utilizing the EpiScreen-assisted workflow significantly outperformed peers who reviewed the same data without AI assistance.

October 8, 2026: The research results were published.

The Tech Race

This development represents a major shift toward integrating large language models into clinical decision-making processes. By automating the analysis of unstructured electronic health records, EpiScreen positions AI as a core partner in traditional medical diagnostics.

For patients, this technology could lead to faster and more accurate epilepsy diagnoses without the need for prolonged hospital stays. For doctors, the AI provides a collaborative tool to reduce errors in complex neurological evaluations.

The takeaway

The success of EpiScreen underscores the potential for AI to bridge critical gaps in medical specialization and resource availability. Patients awaiting neurological assessments may soon benefit from faster turnaround times as hospitals integrate these diagnostic tools.