The artificial intelligence will see you now.
While that might be a bit of an exaggeration in 2023, patients may not be too far off from hearing those words at their physician’s office.
One specialty where AI is becoming increasingly common is gastroenterology, where devices that learn from millions of previous images are being used in colonoscopies to detect polyps that might have otherwise been missed. Health systems including New Hyde Park, N.Y.-based Northwell Health, Salt Lake City-based Intermountain Health, Chicago-based CommonSpirit Health have hospitals employing the new technology.
Health system digital and GI leaders told Becker’s the tool is showing promise and being asked for by some patients, though it’s still in its early stages.
“We are starting to see an increased awareness of the value potential of AI in colon polyp detection that is spreading across our communities,” said Nathan Merriman, MD, medical director of gastroenterology and digestive health for Intermountain Health.
“There have been several patients we have heard who have requested a colonoscopy with the new technology. As our patients and our providers start to hear more about the increase in colon polyps we are seeing with this new technology, I suspect there will be more requests for colonoscopy exams with this new AI teammate in colon polyp detection.”
A March 2022 study in Gastroenterology found that using AI during colonoscopies may help decrease adenoma miss rates. AI-aided colonoscopies saw a twofold reduction in the miss rate of colorectal neoplasia compared to standard colonoscopies, the researchers concluded.
“Despite [colonoscopy’s] obvious advantages, there are still limitations to it,” said Shyam Thakkar, MD, director of advanced therapeutic endoscopy at Morgantown-based West Virginia University School of Medicine, who is developing an interventional endoscopy program for 19 hospitals. “Small polyps or even significantly sized polyps can be missed at the time of the colonoscopy. AI helps us in the sense that while the endoscopist is performing an exam, this miss rate can be reduced because it actually helps us identify where these polys might be.”
The most widely-used AI-powered gastroenterology device is Medtronic’s GI Genius endoscopy module. In February 2022, the medical device company partnered with Amazon Web Services to launch the Health Equity Assistance Program, which donates the devices to facilities in low-income and underserved communities to increase access to AI-assisted colorectal cancer screenings.
Through the initiative, WVU Medicine’s J.W. Ruby Memorial Hospital in Morgantown, University Hospitals Cleveland Medical Center, and Grady Memorial Hospital in Atlanta became the first hospitals in their respective states to use GI Genius.
In addition to GI Genius, the FDA has also approved Wision AI’s EndoScreener and Iterative Health’s SKOUT for AI-assisted polyp detection.
How to temper the hype
At Northwell Health, Michael Oppenheim, MD, senior vice president clinical digital solutions, often has to temper physicians’ enthusiasm about the latest in AI.
“Our job is to help make the decision-making process more scientific, and help them see through the hype,” he said. “Understand the biostatistics, understand the performance characteristics, understand the true value, the potential financial upsides and downsides.”
The health system works closely with device vendors to make sure the products are not only properly validated but constantly updated, Dr. Oppenheim said. “The optics of your colonoscopes may change over time. Is the vendor staying up on that and making sure their algorithms are retrained?” he said.
Northwell also consults with clinical partners that helped develop the tools and relies on feedback from clinicians on the devices’ efficacy, he said.
AI in healthcare is so new that regulations and norms are still emerging, said Nigam Shah, PhD, chief data scientist at Palo Alto, Calif.-based Stanford Health Care.
“Currently, the nature of the device, the kind of decision it will enable, and the degree to which a human is in the loop are the key factors in devising a selection and monitoring strategy,” he said. “We rely on the draft guidance from the FDA and [National Institute of Standards and Technology] in order to inform our strategy. Unfortunately, there is not a one-size-fits-all recommendation that can be made yet.”
Where hospitals can try the tech
When selecting a device, hospitals and health systems can trial it or look at study outcomes, costs, and the company’s history and reputation, said Toufic Kachaamy, MD, chief of medicine and director of gastroenterology and endoscopy with City of Hope in Phoenix.
“My advice, however, is to avoid buying a software and instead subscribe to a software that updates with time,” he said. “I also recommend looking at the outcome they are trying to affect at baseline and to regularly look at the outcome after the introduction of the software.
“If the software delivers on its promise and the metric moves in the right direction, then hospitals can decide the impact of the software and if it makes clinical and financial sense to keep subscribing to the software for their own practice and their patient population.”
So, what will the use of AI in gastroenterology look like in the near future?
“Endoscope manufacturers and other endoscopic technology companies understand the potential of AI,” said Gerard Isenberg, MD, chief medical quality officer for the University Hospitals Digestive Health Institute. “The costs of AI image analysis will continue to fall over time as the technology matures and more competitors enter the marketplace. Physicians and patients will demand that this technology be incorporated for all of their procedures.”