Penn team tests AI chat system that lets patients message implanted brain monitors

Researchers at the University of Pennsylvania have piloted an artificial intelligence platform that lets patients message their implanted brain monitoring devices and receive feedback through a smartphone application, according to Penn Medicine.

The system was tested in the hospital setting: the team enrolled 13 adults who were undergoing continuous electroencephalogram (EEG) monitoring in Penn’s Epilepsy Monitoring Unit and collected interactions between patients and the platform. Penn Medicine reported that investigators analysed 1,307 messages exchanged between patients and the system over the study period.

The platform is described by Penn as an AI-driven model for patient-device communication that enables two-way exchanges of information and real-time feedback. Penn’s account says the program is intended to allow devices and patients to “learn together” and to adapt disease management dynamically rather than rely on static settings.

The research builds on work led by Brian Litt at Penn to create a new generation of brain-computer interfaces that sense physiologic signals, communicate with patients, learn from those responses and adapt treatment in real time.

Penn Medicine states the pilot focused on people in its Epilepsy Monitoring Unit who were already receiving continuous EEG monitoring, and the report gives the numbers of participants and total message exchanges as the primary metrics. The university’s engineering and medical outlets published accounts aligning with Penn Medicine’s description of the project and its aims.

Headlines at other outlets noted the same basic premise: enabling real-time conversation between patients and medical devices using AI and smartphone interfaces. Penn Medicine also published funding details, saying the work is supported by a National Institutes of Health Pioneer Award (DP1 NS122038-01) valued at $5.6 million.

The article ties this pilot into Litt’s broader research program on brain-computer interfaces; Penn frames the study as an early test of systems that can sense physiologic signals and adapt treatment as they learn from patient interactions.

The reporting available is a bounded account drawn from Penn Medicine’s write-up: it provides participant count, the monitoring setting (continuous EEG in the Epilepsy Monitoring Unit), the total number of patient-system message exchanges, and the NIH award supporting the research.

The published material presents those measures as the key outputs of the pilot and situates the project within Litt’s ongoing research agenda at Penn.