During the past year, Southern Illinois University School of Medicine has been quietly refining its latest innovation in medical education: a virtual patient simulation that uses artificial intelligence (AI).
The virtual patient uses AI to help students hone their communication and interpersonal skills in a safe online environment. The program pairs sophisticated rendering technology with real human case studies.
Dr. Richard Selinfreund, associate professor at SIU, spearheaded the project in response to student feedback about outdated virtual training software. He enlisted the help of the medical students, faculty and SIU’s Digital Humanities Lab in Carbondale, led by Professor Pinckney Benedict. Together, the team created Randy Rhodes, a computerized patient capable of realistic, ever-changing interactions.
The patient can simulate various health conditions, personalities and challenging scenarios, which the students navigate through dialogue. As a learner progresses through the training, the complexities will increase. The virtual patient will have confounding comorbidities (a second medical condition that complicates the main disease being studied), mental health issues, and even delve into sensitive topics like suicidal ideation and discrimination.
As the project evolves, it could revolutionize physician training methods across the country and beyond.
"The potential’s practically limitless,” Selinfreund said. "Imagine that every person, every medical expert could train on their phone or on a computer before they walked into a scenario and actually see a real patient. They can develop confidence and get a measurement of their patient’s trust. It’s an enormous breakthrough in medical education.”
The AI project builds on SIU’s rich history of educational innovation and its problem-based learning curriculum. The pioneering use of standardized patients in the 1980s began under Dr. Howard Barrows, creator of the now-common practice of having actors portray illnesses to groups of medical students.
The autonomous AI in the virtual patient allows for spontaneity and adaptability within the simulations, comparable to flesh-and-blood clinical encounters.
"This is really a tool for teaching medical students and resident physicians the kind of soft skills for communication that aren’t prominent anywhere else in the curriculum,” said Tommy Johnston, a fourth-year medical student on the team.
Johnston and other team members have been presenting their work at national conferences, generating significant interest in the medical education community. Six medical schools are beta testing the AI patient program in collaboration with SIU Medicine.