First-of-its-Kind AI Tool for Diabetic Retinopathy Detection Approved by FDA

The FDA today approved what it’s calling “the first medical device to use artificial intelligence to detect greater than a mild level of the eye disease diabetic retinopathy.” The AI-powered, cloud-based system will be available for use by primary care providers.

Over 30 million Americans have diabetes, and diabetic retinopathy—which occurs when blood sugar levels result in damage to retinal blood vessels—is considered mostly preventable. Still, it causes vision loss in tens of thousands of people each year and is the leading cause of blindness among working-age Americans.

“Many patients with diabetes are not adequately screened for diabetic retinopathy since about 50 percent of them do not see their eye doctor on a yearly basis,” Malvina Eydelman, MD, said in the FDA’s official announcement. She serves as director of the Division of Ophthalmic, and Ear, Nose and Throat Devices at the agency's Center for Devices and Radiological Health.

IDx-DR, the flagship technology from Iowa-based company IDx, is cloud-based software designed to detect the condition early on. A primary care doctor can photograph a patient’s eyes using a retinal camera and then upload those images into the software, which will return 1 of 2 possible results: If it detects “more than mild diabetic retinopathy” it recommends referring the patient to an eye care professional; and if the patient is negative for the condition, it recommends rescreening in a year.

In February, the FDA granted IDx-DR an expedited review as a “breakthrough device.” Today’s approval was based on a multisite clinical study that included over 900 patients at 10 test sites. The system was able to identify those with the condition 87.4% of the time, and those who did not 89.5% of the time.

“The healthcare system desperately needs a more efficient and cost-effective way to detect diabetic retinopathy. Too many patients go blind needlessly because they aren’t diagnosed in time,” Michael Abràmoff, MD, PhD, founder and president of IDx, said when the software was granted expedited review. He added that that decision confirmed what he and his colleagues had “believed for a long time.”

Image analysis has, early on, been one of AI’s main strong suits in healthcare. Numerous companies are pursuing diagnostic systems meant to aid primary care physicians and radiologists (or replace that latter, depending on who you ask). Just last month, researchers from Google published a study in the journal Ophthalmology about their efforts to develop a diabetic retinopathy detection system.

With a glut of new AI-based diagnostic and decision support tools knocking at its regulatory door, the FDA has been trying to create streamlined review processes for software and devices. Since assuming the role as the agency’s commissioner, Scott Gottlieb, MD, has released a torrent of new guidance documents and promises of new regulatory pathways.

The agency might have a bit of catching up to do, however. Today’s IDx-DR approval is major news, but clinical practices in Europe have been using the technology since it was approved for sale there in 2013.

“Today’s decision permits the marketing of a novel artificial intelligence technology that can be used in a primary care doctor’s office,” Eydelman said in the FDA’s announcment. “The FDA will continue to facilitate the availability of safe and effective digital health devices that may improve patient access to needed health care.”

Related Coverage:

Employing Eye Experts, Google Sharpens Its Disease-Detecting AI

Another Start-Up Advances AI's Assault on Diabetes

Expedited FDA Review for Novel AI That Detects Diabetic Retinopathy