Penn tests conversational AI for patient–device communication during seizure monitoring
Published in Artificial Intelligence, Brain/Neurology, Personalized medicine, Tools.
Published October 14, 2026.
University of Pennsylvania researchers have tested a smartphone interface that lets patients exchange messages with an AI system connected to their brain-monitoring data. Penn Medicine reported the proof-of-concept study on October 13, with Penn Engineering highlighting it on October 14.
The study involved 13 adults in an epilepsy monitoring unit and analyzed 1,307 messages. Participants could ask about EEG patterns, view trends, report symptoms and receive alerts. Patient and clinician annotations of seizures and false alarms helped refine the system’s detection.
The practical idea is to combine physiological measurements with information that sensors cannot capture, such as a person’s symptoms or daily experience. However, this small inpatient study does not establish that the platform reduces seizures or is safe for unsupervised home use. Effects on seizure severity, quality of life and health-care use remain subjects for future evaluation. Automatically adjusting treatments is a proposed next step, not an established result of this study.
Source: Penn Medicine announcement and linked Nature Medicine study.