Implementing an artificial intelligence tool to help reprioritize radiologists’ CT work lists has helped one Texas hospital reduce turnaround and wait times, experts reported Wednesday.
The machine learning algorithm works by flagging abnormal, noncontrast exams for intracranial hemorrhage. Testing the technology out in its busy academic neuroradiology practice between 2017 and 2019, University of Texas Southwestern Medical Center found marked gain, according to an analysis published in Radiology: Artificial Intelligence. In particular, active reprioritization of the work list with AI dropped wait times from about 15.75 minutes per study down to 12.01.
“We have shown that the major bottleneck for report turnaround time is the amount of time an examination waits in the work list,” Thomas O’Neill, MD, an assistant professor of radiology at UT Southwestern, and colleagues wrote Nov. 18. “AI-based automatic prioritization appears to have significant potential to reduce report turnaround time in targeted populations, due to its effect on wait time,” they added later.
O’Neill et al. put the commercially available algorithm to use in three separate places: as a pop-up widget on ancillary monitors, a marked examination in reading work lists, and a marked examination for reprioritization based on the presence of a flag.
The research team found that notifications with a widget and flagging the exam had no effect on image turnaround or wait times. However, they did score gains with artificial intelligence-detected intracranial hemorrhage cases with reprioritization. And those wait-time improvements persisted across all order classes but was most pronounced for exams ordered as “routine” for inpatients and outpatients, “due to their low priority,” the authors noted.
You can read more about their findings in RSNA’s Radiology: AI journal here.