Northeastern Gets $900,000 to Develop AI Patients for Healthcare Training

Northeastern University researchers have received a three-year, $900,000 National Science Foundation grant to develop virtual patients that healthcare students can practice talking with. The funding will advance the Virtual AI Patient Simulator, an AI-powered virtual reality system already built and piloted, and support tools that let clinical faculty create patients for their own courses.

The award brings together faculty from Northeastern’s arts and design, computer science, and health sciences colleges. It comes through the NSF’s Research on Innovative Technologies for Enhanced Learning program. The collaboration began a couple of years ago with a university seed grant intended to establish proof of concept and help multidisciplinary teams compete for external funding.

Clinical cases become virtual conversations

The headset-based system, known as VAPS, is designed for real-time clinical communication practice. Its virtual patients’ conversational programming is shaped by actual clinical case records, according to principal investigator Eileen McGivney. The team is working on customizable patients with realistic speech, eye contact, facial expressions and gestures—not just a conversation detached from a clinical setting.

McGivney describes the educational challenge as giving students an unscripted clinical environment that is difficult to reproduce outside hospitals and doctors’ offices. Patient encounters vary by setting and circumstance, while opportunities for hands-on experience can be limited. The project is intended to give students more chances to practice before entering those real-world settings.

What excites us about VAPS is that there is an opportunity to give learners safe environments where they can practice navigating challenging conversations, receive feedback, reflect and try again

Heidi Cheerman, associate clinical professor in Northeastern’s department of physical therapy

Faculty-created patients, then performance assessment

The planned authoring tool would let clinical faculty without experience in VR or AI create patients suited to their curricula and classrooms, McGivney said. That puts scenario creation in instructors’ hands rather than requiring them to build the underlying technology. It is a development goal for the grant, not a newly released faculty tool.

Researchers are also using eye-tracking and heart-rate data collected through the headsets to measure students’ responses during interactions. A separate animation system is under development to make patients’ appearance and nonverbal behavior more realistic, said researcher Leanne Chukoskie. These efforts address both how students respond and how the simulated patient presents itself.

Cheerman said the goal includes cultivating communication, empathy and listening, along with helping students adapt to the moment. The team plans further platform testing and assessments to gauge student performance. Sheri Kiami, a physical therapy faculty member with a simulation-education background, is helping incorporate industry best practices, another requirement of the grant.