UofL School of Medicine faculty uses AI to build clinical tool in five weeks

After more than four decades practicing medicine, Charles Barr, professor of ophthalmology, knows the importance of being able to compare a patient’s retinal images from one visit to the next. He also knows what he needs from a tool that can help him do it. 

Earlier this spring, Barr enrolled in MBA 680: AI for Business Innovation, an elective through the University of Louisville’s College of Business. He had genuine skepticism about AI-assisted coding. Five weeks later, with no background in software development, Barr used generative AI to build a browser-based tool that aligns retinal photographs from different patient visits and highlights changes in a lesion. 

The prototype focuses on geographic atrophy, a progressive condition that can cause vision loss and requires clinicians to understand how the disease changes over time. Barr is now in conversations with his department about developing the prototype into a clinical product. 

“I am impressed that AI could build what I wanted in a matter of a few minutes rather than a few weeks,” said Barr. 

But Barr’s experience also demonstrates what AI cannot replace: the human expertise needed to understand a problem, determine what matters and decide whether a solution is truly useful.  

Barr knew which structures in the eye needed to align before an image comparison could be trusted. He knew the central macula was the clinically important area. And he knew that a tool designed for use in an exam room needed to be simple and intuitive. Each of those decisions shaped his prompts to the AI. 

“I am a physician, not an engineer,” said Barr. “All I need is a tool that lets me upload two retinal images side by side.” 

Barr worked through five iterations of the prototype, correcting the AI each time it missed the mark. The technology generated the code while he provided the clinical context and judgement.  

“We’re already seeing AI deeply integrated into mature clinical systems,” shared In Kim, vice dean of strategy and innovation for the School of Medicine. “We’re being deliberate about how we expand AI at the School of Medicine. We want to make sure that as we embrace these powerful tools, we never lose sight of the human elements of medicine – the judgement, relationships and compassion that are at the heart of caring for patients.” 

That balance between technology and human expertise is at the heart of MBA 680. Barr wasn’t the only student without a programming background. Students used AI to develop tools addressing challenges ranging from hearing aid comparisons and wound-healing therapeutics to leadership bias and team building. 

The course was designed and taught by Sarah Bhatia, COB alumna and senior product manager at Slingshot. The course is part of Cardinal Intelligence, UofL’s university-wide initiative to prepare its community to navigate and lead in an AI-enabled world.  

Courses like MBA 680 reflect UofL’s core belief that AI literacy is becoming a professional literacy. Rather than simply discussing what AI might make possible, students are challenged to apply it to real problems in their own fields, building and refining working solutions along the way.  

For Barr, that preparation meant more than learning how to use AI. It gave him the ability to take a problem he had encountered throughout his career and explore a potential solution himself while relying on the knowledge and experience only a clinician could bring to the process.  

As AI continues to evolve, the future of medicine won’t simply be about what technology can do, but how physicians and other health professions use it while applying clinical judgement, human connection and a deeper understanding of the people they serve.