During a Wednesday session, Krishna Nallamshetty, MD, presented the results of a year-long study evaluating the impact of implementing an AI-enabled care coordination program to guide providers and patients during the imaging follow-up process.
Nallamshetty
"For every 1 million diagnostic imaging exams performed, an estimated 135,000 are lost to follow-up," said Dr. Nallamshetty, professor and chair of radiology at the University of South Florida Morsani College of Medicine and chief medical officer of Radiology Partners. "The purpose of our project was to develop a care coordination program to make sure patients get the right study at the right time."
Dr. Nallamshetty said approximately 15 percent of imaging reports include follow-up recommendations. Of those, 66% omit details such as the imaging modality and timeframe required for accurate follow-up. Surprisingly, even when those details are included, the necessary imaging is completed only 30% of the time.
AI Patient Registry Was Integrated Into EMR
In the study, the researchers reviewed imaging studies performed within a 14-month period at a Radiology Partners site using an AI-based natural language processing algorithm to identify patients with incomplete follow-up imaging. Within the study period, the practice also began using another AI program that automatically presents the interpreting radiologist with evidence-based follow-up recommendations based on imaging findings.
Using an AI-driven patient registry to track the status of follow-up recommendations, the care coordination program sent automatic notifications to referring physicians to prompt action on follow-up imaging. If physicians didn't respond, the system sent a notification to the patient to contact their doctor to discuss the recommended follow-up. A team of navigators managed the care coordination program and reached out to physicians and patients directly when needed.
"We aren't telling referring providers how to manage their patients," he said. "Our goal is to be an umbrella that catches any recommendation that falls through the cracks and to empower patients to take care of themselves."
A total of 2,971 studies were identified with significant incidental findings, of which 2,951 required follow-up. The most frequent follow-up recommendations were related to findings in an ovary, kidney or thyroid gland.
At the study's beginning, imaging follow-up recommendations were successfully completed at a rate of 43%. One year later, follow-up exams were successfully completed 91% of the time.
"Using AI programs and care navigators, we increased the number of patients receiving recommended follow-up imaging by nearly 50% or a 110% improvement from baseline," Dr. Nallamshetty said.
Implementation At Other Locations Shows Consistent Results
Following the pilot, the care coordination program has been implemented at additional Radiology Partners sites with consistent results.
"Radiology can provide significant value through a care coordination program by leveraging AI to provide the right study at the right time," he said.
Access the session, " The Right Study at the Right Time - Expanding Radiology's Role in Care Coordination," (T3-SSNPM02-2) on demand at Meeting.RSNA.org.
