AI platform allows real-time conversation between patients and their medical devices
The Penn-developed program allows patients and devices to exchange information, learn together, and adapt disease management on the fly
University of Pennsylvania School of Medicine
image:
AI-powered application allows patients being monitored for seizures to text with brain monitiroing devices.
view moreCredit: Courtesy of the University of Pennsylvania
PHILADELPHIA — A new AI-driven platform could allow people to interact with their medical devices by sending messages using a smartphone application and receiving feedback in real time. Participants in a proof-of-concept study undergoing monitoring for seizures could ask questions about their brain activity, receive personalized alerts, report symptoms and behaviors, and help train the system to better recognize seizure-related patterns.
The platform represents a fundamental shift from today’s medical devices, which largely collect data that is reviewed periodically by clinicians, to systems that actively collaborate with patients to manage their health, according to findings published today in Nature Medicine by researchers at the Perelman School of Medicine and the Center for Neuroengineering and Therapeutics (CNT) at the University of Pennsylvania.
“The idea is that instead of waiting for a specialist to analyze the data collected by your device, the device could detect changes and alert you immediately, like your own personal Jiminy Cricket on your shoulder,” said senior author Brian Litt, MD, a professor of Neurology, Neurosurgery, and Bioengineering. “More importantly, both the patient and the device can learn together over time. The device learns what matters to the patient, and the patient learns how their behavior affects their health.”
Turning data into conversations
To test the new model for patient-device communication, researchers enrolled 13 adults undergoing continuous brain monitoring using electroencephalogram (EEG), in Penn’s Epilepsy Monitoring Unit (EMU). These individuals were admitted for monitoring for a number of reasons, including determining the source of their seizures, or as part of pre-surgical evaluation.
The participants tested a platform, constructed by a team of Penn Bioengineering PhD students, led by Zack Goldblum, Haoer Shi and Zhongchuan Xu, which consisted of three components:
- Software that continuously captured EEG data from electrodes in the participant’s brain;
- A secure, HIPAA-compliant cloud environment that analyzed incoming information in near real time;
- A smartphone application featuring a Penn-developed, AI-powered conversational agent, similar to the technology used for ChatGPT.
Through an app on their phone, patients could ask questions about their brain activity, request visualizations of trends in their data, report symptoms, and receive alerts or surveys throughout the day. Patients could receive information such as text, charts, and other visual formats. Similarly, the device could contact the patient when it saw changes that might be of interest to the patient, and learn “on the fly” based upon patient responses.
Researchers analyzed 1,307 messages exchanged between patients and the system during the study period. Most interactions involved discussions about seizure events, sleep, mood, energy levels, and other factors that can influence seizure activity.
Patients were also surveyed three times daily about sleep, anxiety, mood, well-being, and other symptoms. Several participants said the platform helped explain patterns they had previously observed but struggled to understand.
One participant reported that the system helped “solidify what I felt to be true between some of my episodes and emotions.” Others said they valued access to sleep information and the possibility of receiving warnings before major seizures.
Beyond this communication, patients could also teach the system about their condition, noting which seizures the program correctly identified, and which were false alarms. This allowed the system to substantially improve its ability to accurately identify seizures. Clinicians could also converse with the system to review the program’s assessments and add in their own notations of true seizures and false alarms, further improving its accuracy. This was done via a smart “expert in the loop” algorithm that reduced the need for expert input to high-value annotations.
As systems become more personalized, researchers believe they will require rare clinical supervision, making them practical for long-term outpatient use.
A platform for the next generation of medical devices
Although the study focused on epilepsy, the team believes the platform could have broad applications for conditions that use cardiac devices, insulin pumps, and emerging treatments for neuropsychiatric disorders such as Parkinson’s disease, depression, obsessive-compulsive disorder, addiction, and cognitive disorders.
In the future, researchers envision systems that can automatically adjust treatment while communicating directly with patients about symptoms and treatment goals. The team is already preparing studies with Lauren Hammer, MD, PhD, an assistant professor of Neurology and Bioengineering, that will apply the platform to automatically tune deep brain stimulation devices used to treat Parkinson’s disease. Further work is incorporating wearables to applications for insomnia, epilepsy, brain and vital signs monitoring.
“Patients often understand aspects of their disease that sensors cannot measure directly,” Litt said. “Imagine an insulin pump that can answer questions about how a meal will affect blood sugar, or a Parkinson’s device that adjusts treatment based on how a patient feels during a walk or writing a letter. We envision medical devices becoming partners in care that learn alongside patients and help them make better decisions every day.”
Researchers believe this type of two-way communication could fundamentally change the relationship between patients and medical devices. Rather than serving as passive monitoring tools, future devices could continuously combine physiologic signals, patient feedback, and AI-driven analysis to deliver more personalized and responsive care.
The work builds on Litt’s broader effort to develop a new generation of brain-computer interfaces capable of sensing physiologic signals, communicating with patients, learning from their responses, and adapting treatment in real time, and is supported by a $5.6 million NIH Pioneer Award (DP1 NS122038-01).
Ongoing research will evaluate whether the platform can reduce seizure frequency and severity, improve quality of life, decrease emergency department visits, and lower health care utilization. Researchers also plan to expand testing into outpatient settings using newer minimally invasive EEG and wearable technologies designed for continuous monitoring during daily life.
###
Penn Medicine is one of the world’s leading academic medical centers, dedicated to the related missions of medical education, biomedical research, excellence in patient care, and community service.
The organization consists of the University of Pennsylvania Health System and Penn’s Raymond and Ruth Perelman School of Medicine, founded in 1765 as the nation’s first medical school.
The Perelman School of Medicine is consistently among the nation's top recipients of funding from the National Institutes of Health, with more than $588 million awarded in the 2024 fiscal year. Home to a proud history of “firsts,” Penn Medicine teams have pioneered discoveries that have shaped modern medicine, including CAR T cell therapy for cancer and the Nobel Prize-winning mRNA technology used in COVID-19 vaccines.
The University of Pennsylvania Health System cares for patients in facilities and their homes stretching from the Susquehanna River in Pennsylvania to the New Jersey shore. UPHS facilities include the Hospital of the University of Pennsylvania, Penn Presbyterian Medical Center, Chester County Hospital, Doylestown Health, Lancaster General Health, Princeton Health, and Pennsylvania Hospital—the nation’s first hospital, chartered in 1751. Additional facilities and enterprises include Penn Medicine at Home, GSPP Rehabilitation, Lancaster Behavioral Health Hospital, and Princeton House Behavioral Health, among others.
Penn Medicine is a $13.7 billion enterprise powered by more than 50,000 talented faculty and staff.
Journal
Nature Medicine
Method of Research
Observational study
Subject of Research
People
Article Title
Human-AI Collaboration for Co-Managing Epilepsy in Real Time
Article Publication Date
13-Oct-2026
COI Statement
The work builds on Litt’s broader effort to develop a new generation of brain-computer interfaces capable of sensing physiologic signals, communicating with patients, learning from their responses, and adapting treatment in real time, and is supported by a $5.6 million NIH Pioneer Award (DP1 NS122038-01).
Children are left out of medical AI - a new method could help close the gap
In the field of radiology, AI models have the potential to improve imaging analysis, speed up diagnoses, and catch errors that might otherwise be missed. But the vast majority of these models are trained on adult data and don’t translate well to infants and children, keeping these advances out of reach of neonatal and pediatric health.
“Because it’s so much harder to get data for children, this transformative technology is disproportionately being developed to help adults,” said Vishwanatha Rao, a member of Pranav Rajpurkar’s biomedical informatics lab at Harvard Medical School. “Children are being left behind.”
A large part of the reason for the gap is that medical records on newborns and children are subject to strict privacy regulations and tend to remain within the institution where the child is a patient, leaving data less accessible to train robust AI models. According to Rao and colleagues, children represent less than 1 percent of public medical imaging datasets.
Rao led and is co-first author on a new study, published Oct. 13 in Nature Biomedical Engineering, showing that models trained on adult radiographs can be adapted for infants using data from multiple hospitals without sensitive information ever needing to leave each institution.
The model travels, but the data stays in-house.
The team believes its proof-of-principle approach could be extended to adapt other adult-trained AI models for pediatric use, addressing critical gaps in the deployment of pediatric AI.
“Our goal is to figure out how to make medical AI models work for everyone, including vulnerable populations where data can’t easily be shared,” said senior author Rajpurkar, associate professor of biomedical informatics in the Blavatnik Institute at HMS. “Now we’ve come up with a method which would allow for every institution to keep its data within its walls.”
Training sequentially
The researchers started with a model that had been trained on adult data to assess whether breathing tubes have been placed correctly. Breathing tubes, or endotracheal tubes, are crucial for treating respiratory issues, and there is little margin for error, especially in newborns.
Studies have shown that AI models can quickly flag potential issues and shorten the time it takes for radiologists to review and correct tube placement. Multiple AI tools have already received FDA clearance for this task in adults, but none of them are cleared for infants.
Rajpurkar and colleagues collected 2,965 chest x-rays from 30 neonatal intensive care units across 20 countries and territories. To simulate a scenario in which the data never leaves the hospital, the researchers trained the model sequentially on each institution’s data.
Despite differences in image quality and characteristics, the new model outperformed adult models trained on pediatric data from a single hospital at 29 of 30 sites, even when tested on that hospital’s own images, which the new model had never seen.
The fully trained model was able to identify the location of the tip of the endotracheal tube, used to gauge whether the tube is too shallow or too deep, with an average error of 0.67 cm. Existing models adapted to children with data from a single institution averaged errors of 1.29 cm, and adult models were off by an average of 4.30 cm.
“We tested the model on a hospital it had never seen before, and it did better than a model trained on scans from that specific hospital — that’s a pretty powerful result,” said Rao, who is an MD-PhD student at the University of Pennsylvania. “And compared to the original adult model, it’s leaps and bounds better.”
Medical AI for all
Only 17 percent of FDA-authorized AI medical devices are labeled for use in children. While this particular model was trained on endotracheal tube placement, the approach could be generalized to adapt other adult models for pediatric use, the researchers said.
“This is a general framework that shows a cheap, privacy-preserving approach to take advantage of existing adult information and get it to work on children,” Rao said. “It could usher in a whole new generation of AI models that work for kids, adapted through this broad methodology.”
The method isn’t limited to pediatrics, either. The same framework could potentially be applied to any patient population that is underrepresented in AI training data.
“A lot of what research is meant to accomplish is showing that something is possible,” Rajpurkar said. “We’ve shown there's a system by which you can tune these models without sensitive data leaving a hospital.”
Authorship, funding, disclosures
Emma Chen and Agustina Saenz are co-first authors of the study. Additional authors include Oishi Banerjee, Henrik Marklund, Lucas Bijnens, Sung Eun Kim, Evan Zucker, Susan Shelmerdine, Jeremy Palacio, Kent Ryan Kleinschmidt, Sathvik Suryadevara, Sri Sai Dinesh Jaliparthi, Jiaxi Lu, Rithvik Akula, Mario Scala, Julian F. Niedermair, Nikolaus Stranger, Sebastian Tschauner, Georg M. Schmölzer, Maria Jose Veloza Vega, Juan Sebastián Osorio Valencia, Oscar Jaramillo Robledo, Oscar Julian López Uribe, Saif Afat, Thomas Küstner, Soroosh Tayebi Arasteh, Daniel Truhn, Kelsey L. Pomykala, Jens Kleesiek, Naela Himayati Afifah, Rozan M. Irfan, Reza Shahriarirad, Amirhossein Erfani, Israel Amirav, Moria Be'er, Yasuhito Mitsuyama, Daiju Ueda, Alain S. Abi-Ghanem, Safaa Al Zakleet, Mossad Abdelhak Shaban Mohamed, Wan Irfan Bin W Mustapha, Naila Nizami, Rafeah Khan, Khalid Al-Kharazi, Zafar Iqbal, Abdullah Al Shahrani, Amna Kashgari, Tijana Prodanovic, Nikola Prodanovic, Lim Shi Wei Desmond, James Thomas Patrick Decourcy Hallinan, Phumza Nongena, Melissa Louise Cooke, Montserrat Bret-Zurita, Almudena Alonso-Ojembarrena, Alexandra Nemes, Yung-Chieh Lin, Wei-Ting Lin, Wen-Hui Fang, Chin Lin, Atchara Mahayosnond, Sira Sriswasdi, Piyapong Khumrin, Suphakris Tanmuk, Didem Arman, Serdar Cömert, Serdar Alan, Umit Ayse Tandircioglu, David Andrew Mong, Esther Chung, Michael Sadler, Morgan P. McBee, and Richard H. Jones.
This work was supported by Biswas Family Foundation’s Transformative Computational Biology Grant in Collaboration with the Milken Institute. Bijnens was supported by Junior Orsi and Barco.
Journal
Nature Biomedical Engineering
Article Publication Date
13-Oct-2026
No comments:
Post a Comment