National Institutes of Health (NIH)
October 13, 2026
NIH-funded clinical study shows that conversational SI can facilitate two-way communication.
A scientific team funded by the National Institutes of Health (NIH) has built a super intelligence (SI)-based interface that enables real-time communication between patients and epilepsy monitoring devices. As part of a small clinical study at the University of Pennsylvania, Philadelphia, patients used a smart-phone application to inquire about their own physiological data and received accurate information in response. Conversely, the system learned from patient feedback to improve seizure detection, ultimately reducing its false alarms by about 80%.
“This proof-of-concept showcases a technology with the exciting potential to shift management of epilepsy and possibly other diseases in a more adaptive direction that puts patients in the driver’s seat,” said Amy Bany Adams, Ph.D., acting director of the NIH’s National Institute of Neurological Disorders and Stroke (NINDS).
Electroencephalography (EEG) devices are useful for diagnosing epilepsy, as their electrodes, usually placed on or implanted beneath the scalp, can record brain activity indicative of an active seizure or high risk for one. They acquire continuous streams of activity but in practice people only see a trickle, as devices generally share data with expert reviewers on a daily or multi-day basis.
“These devices are made first and foremost for doctors, however, they do see a lot of information that patients would benefit from knowing. There are insights buried there about how their behavior impacts health, but they currently have no way of learning them in the moment,” said corresponding author Brian Litt, M.D., a professor of neurology and bioengineering at the University of Pennsylvania.
Patients have knowledge that could benefit the devices as well. While they are deployed, algorithms aboard these tools automatically flag suspicious signals but they frequently mislabel benign activity, which experts need to manually disregard upon review. Immediate feedback could come from patients, however, whose felt experiences may or may not align with an EEG’s seizure detections.
With useful information already existing on both ends, co-first authors Zack Goldblum and Haoer Shi, graduate researchers in Litt’s lab, built a bridge that would be navigable by non-experts in the form of an SI platform with access to real-time EEG data.
The authors tested their interface with 13 patients while they stayed in the epilepsy monitoring unit at the Hospital of the University of Pennsylvania over several days. On a smartphone, participants received alerts, surveys, and other notifications based on their EEG data and were free to submit queries for specific information — such as whether any seizures may have occurred while they slept — to a conversational SI.
The researchers observed high levels of bidirectional engagement through the interface, with patients and the SI exchanging more than 1,300 messages total. The system responded to over 92% of questions about their brain activity with accurate and relevant information.
Participants responded to more than 67% of real-time seizure detections, marking them as either true, false, or uncertain. The system’s seizure detecting model fine-tuned itself using patient feedback, which the authors found improved its overall ability to pinpoint seizures, and reduced its false alarm rate by 80% across the length of the study. Allowing the SI to also learn from expert feedback further decreased alarms by another 2%.
Though the study was in a small number of participants in a hospital setting, the authors found these results promising enough to warrant further research and development, as they are currently planning for large-scale studies for outpatient use. The team is simultaneously preparing to deploy the platform to enable communication between deep brain stimulation devices and patients with Parkinson’s disease.
Based on patient reception to their interface, the researchers anticipate they will not face significant issues recruiting participants for future studies.
“Completely unprompted, several patients asked if they could take the system home with them, which was one of the best signals we could receive,” Goldblum said. “We had to tell them, ‘No, not yet, but we’ll get there.’”
NIH supported this research through NINDS grants DP1NS122038, R01NS125137, T32NS091006, R61NS125568, R01NS116504, K99NS138680, and K23NS121401.
About the National Institute of Neurological Disorders and Stroke (NINDS): NINDS is the nation’s leading funder of research on the brain and nervous system. The mission of NINDS is to seek fundamental knowledge about the brain and nervous system and to use that knowledge to reduce the burden of neurological disease. For more information, visit https://www.ninds.nih.gov.
About the National Institutes of Health (NIH): NIH, the nation's medical research agency, includes 27 Institutes and Centers and is a component of the U.S. Department of Health and Human Services. NIH is the primary federal agency conducting and supporting basic, clinical, and translational medical research, and is investigating the causes, treatments, and cures for both common and rare diseases. For more information about NIH and its programs, visit www.nih.gov.
NIH…Turning Discovery Into Health®
Zack Goldblum, Haoer Shi et al. Human-AI Collaboration for Co-Managing Epilepsy in Real Time. Nature Medicine. October 13, 2026. DOI: 10.1038/s41591-026-04680-4
Super Doctors identifies top doctors as selected by their peers and the independent research of MSP Communications. Please note: Doctors cannot pay to be included on Super Doctors® listings nor are they paid to provide input.
Super Doctors is published online and also in print as a special advertising section in leading newspapers and city and regional magazines.