Widespread AI adoption in private practice produces measurable efficiency gains
Widespread artificial intelligence adoption in a private radiology practice can produce measurable efficiency gains, according to a new analysis published Tuesday.
Previous research has demonstrated the utility of AI in narrower contexts, focused on single solutions or clinical concerns. European imaging experts aimed to execute a broader study, involving 10 different AI tools from seven separate vendors.
Their investigation, published in JACR, assessed the impact on workflows and radiologist sentiment across a nearly 5-year implementation period, touching 20 outpatient imaging centers. They found clear benefit, with statistically significant efficiency gains and widespread radiologist adoption.
“Multi-vendor AI at scale was associated with measurable [turnaround time] gains in high-volume modalities,” lead author Dr. Sergey Morozov, founder of Belgium-based consulting firm Medlogic, and colleagues wrote Sept. 29. “Infrastructure latency, not algorithm speed, was the primary barrier to clinical utility,” they added.
The investigation involved three retrospective study cohorts. They included a “technical” one, involving 97,000 exams from a two-year period, used to assess PACS-to-PACS latency and alignment of the 10 AI tools. A second turnaround-time cohort included 21,000 exams from 2025, comparing TAT between AI-available and concurrent non-AI workflows. And finally, a third survey cohort included 58 radiologists, working across the 20-location 3R Swiss Imaging Network based in Sion, Switzerland.
Active AI adoption reached about 91% of physicians, with 66% (or 35/53) reporting regular use of the technology. Median total latency—or the delay in the AI results arriving to rads—was about 2.o6 minutes, about 72% of which was attributable to data routing. The “too late rate”—when AI results reached a radiologist after a report was finalized—was about 7.2% overall. This ranged from about 3% for knee MRI up to 13% for chest CT.
After adjusting for various mitigating radiologist factors such as experience, AI availability was associated with lower median turnaround times for trauma radiography (-26%) and knee MRI (-18%). However, brain volumetry MRI showed no significant change (+9.2%). In further exploratory analysis, net promoter scores declined for chest CT (+38 to −3) and aorta CT (+22 to −25), which the authors deemed as “nominally significant before multiple-comparison correction.”
At about 23% to 25% annualized cost of one radiologist full-time equivalent salary, the program generated 0.69 FTE of capacity through trauma radiography alone (or 0.46 FTE after a radiologist-adjusted sensitivity analysis).
Read more, including potential study limitations, in the Journal of the American College of Radiology.
Marty Stempniak has covered healthcare since 2012, with his byline appearing in the American Hospital Association's member magazine, Modern Healthcare and McKnight's. Prior to that, he wrote about village government and local business for his hometown newspaper in Oak Park, Illinois. He won a Peter Lisagor and Gold EXCEL awards in 2017 for his coverage of the opioid epidemic.