AI-based free-text image ordering may save federal CDS mandate from overburdening providers
Providers favor ordering imaging via a new artificial intelligence-based free-text system compared to using structured direct-search tools, according to new research published in the Journal of the American College of Radiology.
The 2014 Protecting Access to Medicare Act is slated to kick-start in January 2023, and mandates physicians consult clinical decision support systems before ordering advanced imaging exams. But many organizations are reporting roadblocks, including providers complaining structured order entry is both time consuming and limiting.
University of Virginia Health System piloted an AI tool that analyzes providers’ free-text reasons for ordering exams and matches them with structured indications. During the first three months, users opted for the AI tool nearly 60% of the time, authors reported Thursday.
The streamlined tool could prevent PAMA from becoming another "burdensome regulation," the group added.
“Our institution’s early experience with a commercially available AI tool to facilitate structured indication selection through free-text order entry suggests that this approach may lessen the burden on ordering providers and encourage overall institutional success regarding compliance with PAMA,” David S. Gish, MD, with the Charlottesville-based Department of Radiology and Medical Imaging, and co-authors wrote.
For their study, the team looked back on more than 40,000 advanced imaging orders—CT, MRI, PET, etc.—entered during the first three months after implementation (February-March 2020).
Clinicians chose free-text AI for 58.9% of orders compared to directly searching for structured exam indications in 41.1% of cases. The tool also predicted structured indications in 92% of orders with free text.
Gish et al. did find exams ordered via AI were more likely to result in no CDS score (PAMA-noncompliant) compared to the traditional approach. But 35% of these orders were scored because users were allowed to exit the workflow, which didn’t pose an opportunity to score appropriateness. This can be solved by not allowing providers to opt out of the CDS but still finish their order.
The group believes further refinements can produce an even better PAMA-compliant system.
“An opportunity for improvement is an AI tool capable of considering additional patient-specific data from the EHR when determining appropriateness of imaging orders,” Gish and co-authors wrote. “For example, an order in a patient with an MRI-incompatible pacemaker could automatically receive a low appropriateness score or an alert advising against MRI for that particular patient.”
Read the full study here.
Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.