Cincinnati-based Christ Hospital Health Network is one of the first health systems to adopt an AI platform from Epic that has helped detect lung cancer in several patients.
The application uses natural language processing to extract mentions of lung nodules from radiologist reports for lung navigators to review and track. Since mid-August, when the technology went live, the AI has found over 4,200 lung nodule mentions that led to the detection of nearly 100 cancers and the initiation of more than 50 cancer treatments.
"This is quite a significant deal," Christ Hospital Chief Medical Officer Marcus Romanello, MD, told Becker's. "This is multiple lives saved. If you look at the treatment for Stage 1 cancer versus Stage 3 or 4, the outcomes are vastly different."
The health system has increased the number of lung nodules it is tracking sixfold, and the AI program was free to implement as part of an Epic EHR update.
"We were talking about this at our executive team meeting: the continual cost pressures and reimbursement challenges. How do we do more with less?" said Joy Oh, chief information and digital transformation officer of Christ Hospital Health Network. "This is saving time all around in addition to saving lives. Let's have technology work for us, not the other way around."
Christ Hospital has helped make improvements to the model, including an update that has cut false positives, said Christ Hospital clinical systems analyst David Moeller. The positive predictive value has been 93%. The next step is for the technology to note the size, composition and location of the nodules.
"We're asking more and more front-line physicians every day to process and compute information," Dr. Romanello said. "So they're receiving a CT report on the patient, then having to read the entire body of it and make multiple decision analyses of the individual nodules. Let's automate that. Let's make our physicians superhuman in their ability to manage the number of patients under their purview."
Epic is working on a similar tool for abdominal aortic aneurysms.
"It's the same concept of searching through all this text for key insights," Ms. Oh said. "You could apply that to almost any specialty in healthcare where you're looking for something very specific."