An AI tool in Epic is helping detect lung cancer early — and could expand to more use cases.
Cincinnati-based Christ Hospital Health Network continues to see results from the technology since launching it in August. The AI Extracted Findings dashboard has identified cancer in more than 185 patients that might not have otherwise been caught, beginning treatment on 92 of them.
The EHR tool uses natural language processing to scan radiology reports for mentions of lung nodules then notifies lung navigators to follow up with patients.
“Not only are we capturing more incidental lung nodules now that we’re able to extract this data from our radiologist reports, it also helps eliminate or reduce human error,” Christ Hospital lung program nurse practitioner Ashley Campbell, MSN, APRN, told Becker’s. “Obviously, the biggest benefit here is that we are capturing lung cancers at the earliest possible stage.”
In the past, radiologists would have to choose a field in Epic to indicate a finding of a lung nodule. Now the AI does it automatically.
“The issue we had was for incidental findings,” said Christ Hospital clinical systems analyst David Moeller. “There is such a broad range of procedures, and it was difficult for us to require [radiologists] to say yay or nay about a lung nodule for every chest imaging procedure they would review.”
A patient might, say, have presented to the emergency department with suspicion of a pulmonary embolism, a life-threatening condition. When the CT scan came back negative, but also showed lung nodules, those might have gone unreported.
Christ Hospital Health Network is one of the first of a handful of health systems to deploy the technology. The organization has had a 69% early-stage cancer diagnosis rate since August, compared to the national average of 46%. Over 7,000 lung nodules have been detected in that time.
“It’s important to note that six times more of the nodules are being found as a part of this AI program, so we’re catching more patients with nodules that wouldn’t have been found,” said Christ Hospital lung nurse navigator Krystal Asher, RN.
Christ Hospital has been monitoring the AI to make and suggest changes along the way. The health system is waiting on an update from Epic that would prioritize higher-risk nodules. Epic is also developing natural language processing for abdominal aortic aneurysms.
“There is tons of potential for so many different things: esophageal cancer, breast cancer, pancreatic cancer — there are so many subtle findings on CT scans,” said Julian Gutron, MD, medical director of general thoracic surgery at Christ Hospital. “On lung cancer scans, we have incidentals, like from the thyroid, goiter, aneurysms of the aorta. AI could potentially be employed to comb through those scans for other non-lung-related diagnoses.”
“Over the past five years, nationwide, we’ve noted a decrease in the mortality of cancer for all comers,” he added. “And I would dare say that’s in great part because of the dedicated lung nodule and screening clinics like we have here, further enhanced with AI capabilities.”
