Real-world use of generative AI boosts radiologist productivity by up to 40%
A “first of its kind” study published Thursday details how one health system is using generative AI in real-world scenarios to boost radiologist productivity by upward of 40%.
Experts with Chicago-based Northwestern Medicine developed the AI system in-house, allowing them to instantly draft near-complete, personalized radiology reports from X-rays. They tested it on 12,000 X-ray interpretations in live clinical care, with the system producing a 15.5% documentation efficiency improvement and no negative impact on clinical accuracy or report quality.
Researchers touted the model’s “holistic” nature, analyzing entire images for all clinical concerns, according to findings published in JAMA Network Open. It then can generate a report that is 95% complete, personalized to each patient and in each individual radiologist’s reporting style—which they can then choose to review, accept and finalize.
“For me and my colleagues, it’s not an exaggeration to say that it doubled our efficiency. It’s such a tremendous advantage and force multiplier,” study co-author Samir F. Abboud, MD, chief of emergency radiology at Northwestern Medicine, said in an announcement from the 11-hospital system shared June 5.
Instead of adapting existing models such as ChatGPT, Northwestern engineers built the system “from scratch” using data gathered at their organization. This allowed them to create a “lightweight, nimble” AI model tailored to radiology. Engineering experts are embedded within the care team at Northwestern to work on AI tools alongside clinicians. Those involved believe their results prove providers can create real-world AI tools without spending big. Scientists have scored two patents for the technology, which is in the early stages of commercialization.
“Our study shows that building custom AI models is well within reach of a typical health system, without reliance on expensive and opaque third-party tools like ChatGPT,” said co-author Mozziyar Etemadi, MD, PhD, an assistant professor of anesthesiology and biomedical engineering. “We believe that this democratization of access to AI is the key to drive adoption worldwide.”
Researchers conducted the prospective study over a five-month period ending in April 2024. The sample included nearly 24,000 radiographs, with half analyzed using generative AI and the rest without. Interpretations with model assistance took about 160 seconds on average, a 15.5% documentation-time improvement over those that didn’t use AI (at 189 seconds). Peer review of 800 of the exams showed no difference in clinical accuracy nor textual quality. Moreover, the AI model flagged studies containing clinically significant, unexpected cases of collapsed lung with a sensitivity of 72.7% and specificity of 99.9% among almost 98,000 studies screened.
Follow-up research that is not yet published showed that some radiologists saw gains as high as 80%, with the software also effective at examining CT scans. Those involved claim this is the first generative AI radiology tool in the world to be integrated into clinical workflows. They also believe this is the first time such a model has demonstrated high accuracy and efficiency across all types of X-rays, “from skulls to toes.”
As the AI system drafts reports for each image, automation monitors for other critical findings, cross-checking them with patient records. When it pinpoints a new condition that requires urgent intervention, it immediately alerts radiologists. Northwestern is now adapting the technology to detect missed or delayed diagnoses such as lung cancer, according to the announcement.
Experts believe this could serve as a key intervention to aid in the radiologist shortage but emphasized they are not looking to replace human physicians.
“You still need a radiologist as the gold standard,” Abboud said. “Medicine changes constantly—new drugs, new devices, new diagnoses—and we have to make sure the AI keeps up. Our role becomes ensuring every interpretation is right for the patient.”
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.