Using AI to Help Identify Patient Transportation Needs

aerial photo of buildings and roads

As Director of the Healthy Vision Lab, Dr. Andrew Williams, Assistant Professor of Ophthalmology at the University of Pittsburgh School of Medicine, is identifying patients’ transportation needs from existing data in the electronic health record using natural language processing (NLP), a form of artificial intelligence.

“Transportation needs are a barrier to care, and thanks to the Department of Ophthalmology’s patient navigator program, a fairly addressable one,” Dr. Williams said. “Most of the transportation referrals to our patient navigator are successfully resolved by connecting patients to county or insurance-based resources. It seems, therefore, that the limiting factor in addressing transportation needs is identifying patients who have them. We have had modest success with a screening survey for social needs, but the survey uptake is suboptimal. Therefore, I sought a new collaboration to take a different approach.”

By connecting with Yanshan Wang, PhD, a health informatics researcher, Dr. Williams was able to extract free text from ophthalmology clinical notes and search for transportation needs using an NLP algorithm that they iteratively developed. They found that the algorithm was highly accurate compared to manual review. Almost one in a hundred patients had a transportation need documented in their ophthalmology notes at least once.

“Given how many patients we see, I think that is quite a lot,” Dr. Williams said. “And that is only from the ophthalmology documentation!”

Dr. Williams hopes to apply this algorithm within the electronic health record so that it can someday be used to “flag” patients with upcoming appointments who have language suggestive of transportation needs as identified by NLP. Then the Department could perhaps reach out to these patients in advance of a scheduled appointment for a highly targeted screening for transportation needs assessment.

“That is the dream anyway,” Dr. Williams said. “We will start to apply similar algorithms to other needs domains using the dataset. As AI upends everyday practices with its advanced capabilities, I hope to use these advancements to help our most vulnerable patients.”

This project is funded by the David L. Epstein Award, Chandler Grant Glaucoma Society, with support from the Henry L. Hillman Foundation and the Eye & Ear Foundation.