A new Conversation Between Patients and medical devices could let people ask questions about health data and receive real-time feedback. Researchers at the University of Pennsylvania tested the approach with adults undergoing continuous brain monitoring for seizures.
The platform combined software that captured electroencephalogram (EEG) data, a secure cloud environment that analyzed incoming information, and a smartphone application with an AI-powered conversational agent. The team described the work in findings published in Nature Medicine.
During the study, 13 adults in Penn’s Epilepsy Monitoring Unit used the system while receiving continuous brain monitoring. Their admissions involved several needs, including identifying the source of seizures and pre-surgical evaluation.
Conversation Between Patients and Devices
Participants could ask about brain activity, request visualizations of data trends, report symptoms and receive alerts or surveys. The app delivered information as text, charts and other visual formats. The device could also contact a patient when it detected changes that might interest them.
Researchers analyzed 1,307 messages exchanged during the study. Most concerned seizure events, sleep, mood, energy levels and other factors that can influence seizure activity. Participants also received surveys three times daily about sleep, anxiety, mood, well-being and other symptoms.
Several participants said the platform helped explain patterns they had noticed but struggled to understand. One said it helped “solidify what I felt to be true” about connections between episodes and emotions. Others valued sleep information and the possibility of warnings before major seizures.
The system also let patients identify seizures it recognized correctly and flag false alarms. Clinicians could review assessments and add their own annotations. The researchers said this expert-in-the-loop process focused clinical input on higher-value annotations and helped improve the system’s seizure identification.
From Monitoring to Two-Way Communication
Most medical devices collect information for clinicians to review periodically, the researchers said. Their platform instead combined physiological signals, patient feedback and AI analysis in near real time. Researchers described a model in which patients and devices could learn from one another over time.
The project team included Penn Bioengineering doctoral students Zack Goldblum, Haoer Shi and Zhongchuan Xu. Senior author Brian Litt, a professor of Neurology, Neurosurgery and Bioengineering, said patients can understand aspects of their disease that sensors cannot measure directly.
The study focused on epilepsy, but the researchers said the approach could apply to cardiac devices, insulin pumps and emerging treatments for Parkinson’s disease, depression, obsessive-compulsive disorder, addiction and cognitive disorders. They envision future systems that adjust treatment while communicating with patients about symptoms and goals.
The team is preparing studies with Lauren Hammer, an assistant professor of Neurology and Bioengineering, to apply the platform to deep brain stimulation devices used to treat Parkinson’s disease. Other work is adding wearables for insomnia, epilepsy, brain and vital-sign monitoring.
The research has not yet established whether the approach reduces seizure frequency or severity, improves quality of life, or lowers emergency visits and health care use. Researchers plan to evaluate those outcomes and expand testing to outpatient settings using newer minimally invasive EEG and wearable technologies.
For more on related approaches, Meditech Today has covered AI and patient data communication, AI in patient monitoring, and personalized digital health.
The work builds on Litt’s effort to develop brain-computer interfaces that sense physiological signals, communicate with patients, learn from responses and adapt treatment in real time. It is supported by a $5.6 million NIH Pioneer Award (DP1 NS122038-01).
