Researchers at the University of Pennsylvania have developed an artificial intelligence platform that allows people with epilepsy to communicate with their brain-monitoring devices through a smartphone app. The system can answer questions about brain activity, provide personalized alerts and learn from patient feedback to improve seizure detection.
The findings, published in Nature Medicine on October 13, 2026, point toward a new approach to medical technology in which devices do more than collect health data: they can also communicate with patients and adapt to their needs.
How the AI-powered brain-monitoring system works
The platform connects electroencephalography (EEG) equipment, which records electrical activity in the brain, to an AI-powered conversational interface on a smartphone.
During the study, participants could ask questions about their brain activity, review trends in their data, report symptoms and receive notifications based on information collected by their monitoring equipment.
The system could also ask patients about their experiences and use their responses to help interpret possible seizure events. This two-way communication allows patients to contribute information that brain sensors alone may not capture.
For example, a patient could report symptoms or clarify whether a suspected seizure alert matched their experience. The system could then use that feedback to refine its seizure-detection algorithms.
AI reduced false alarms by 80%
In a proof-of-concept study involving 13 adults undergoing epilepsy monitoring, researchers found that patient feedback helped improve the system’s ability to distinguish seizures from other brain activity.
According to the US National Institutes of Health, the seizure-detection model reduced its false-alarm rate by approximately 80% over the study period. Additional expert feedback reduced alarms further.
The system also responded accurately and relevantly to more than 92% of patients’ questions about their brain activity.
These results are promising, but they require careful interpretation. The study involved a small group of patients in a hospital setting, and the reported improvements were evaluated within the research setup. Larger studies will be needed to determine how well the approach works during everyday life.
Why this could change medical technology
Many medical devices continuously collect information but provide patients with limited immediate insight into what the data means. Clinicians may need to review the information separately, leaving patients waiting for explanations or relying on their own observations.
A conversational AI interface could make some of that information easier to understand while allowing patients to report symptoms and experiences as they occur.
The researchers believe the approach could eventually be adapted for other medical technologies, including insulin pumps, cardiac monitoring devices and systems used to treat Parkinson’s disease.
In these applications, patient feedback could help medical devices better account for individual symptoms, behaviors and treatment responses. However, any system used to guide treatment would require rigorous testing, appropriate clinical oversight and safeguards against incorrect alerts or recommendations.
What happens next?
The research team plans to investigate the platform in larger studies and explore applications outside the hospital, where patients manage their conditions during everyday activities.
Researchers are also preparing work involving deep brain stimulation devices used to treat Parkinson’s disease. The longer-term goal is to develop medical technologies that can combine physiological measurements, patient feedback and AI analysis to support more personalized care.
The platform is not yet an established consumer product for people with epilepsy to use independently at home. Its clinical benefits, safety and effectiveness in real-world settings still need to be demonstrated.
If future trials confirm the early findings, conversational AI could help make medical devices more responsive to patients while giving clinicians additional information about how health conditions affect daily life.
Source: University of Pennsylvania, US National Institutes of Health and Nature Medicine.
