HCP Live
Contagion LiveCGT LiveNeurology LiveHCP LiveOncology LiveContemporary PediatricsContemporary OBGYNEndocrinology NetworkPractical CardiologyRheumatology Netowrk

Steven Yeh, MD: Artificial Intelligence Improves Diabetic Retinopathy Detection

Dr. Yeh discusses how Innovations in AI may improve access for patients in resource limited setting and improve detection of vision-threatening diseases.

Artificial intelligence in the field of ophthalmology may provide an avenue forward to both increase accessbility and also better inform care.

In an interview with HCPLive at the American Academy of Ophthalmology 2022 Meeting, Steven Yeh, MD, Director of Retina and Uveitis Service, Truhlsen Eye Center, University of Nebraska, discussed innovations in AI to detect vision-threatening diseases, including the use of the FDA-approved AI EyeArt technology.

When looking at diabetes as a public health crisis, Yeh noted it is difficult to conceptualize or reach the patient populations that might have diabetic retinopathy that may lead to vision loss.

Thus, the use of AI can provide different avenues to detect vision-threatening diabetic retinopathy so that patients can be properly referred in a timely manner.

"I think ultimately we will use innovation to improve access for many of our patients in resource limited settings, both within the United States and hopefully abroad as well," Yeh said.

For EyeArt specifically, Yeh highlighted its ease of use and ability to generate a report that makes it easy for a primary care provider to refer the patients. He added that the sensitivity is high and the specificity for diabetic retinopathy allows patients to be referred in a timely manner.

He compared to use of AI to traditional models, like one-on-one physician interaction, or new innovations like telemedicine screening, where an eye care provider is required to read the fundus photography. For AI, the algorithim provides immediate feedback and may lessen the gap in patients who may get lost to follow-up.

Yeh added that the amount of data collected may ultimately determine proper implementation strategy.

"It really is important to very thoughtfully map out after a certain number of patients will be screened, what's the percentage likelihood that individuals will be identified who have vision threatening disease and have that access point," he said. "So, they know where to go as a next step just to avoid the the nuances and the problems that may arise if the system isn't ready to bear these types of volumes."