FDA Workshop Tackles Key Questions on Digital Ophthalmic Devices

The 21st Century Cures Act clarified the FDA’s regulation of medical software, amending the definition of "device" in the Food, Drug and Cosmetic Act to exclude certain software functions. The FDA is currently developing draft guidance for public comment to help industry and FDA staff understand how the 21st Century Cures Act affects FDA's oversight of medical device software (more information:

As part of the effort to inform this initiative, the US Food and Drug Administration (FDA) Center for Devices and Radiological Health (CDRH), along with 5 cosponsor organizations, hosted the first Ophthalmic Digital Health Workshop in October. Cosponsoring organizations were the American Academy of Ophthalmology (AAO), the American Academy of Pediatrics (AAP), the American Association for Pediatric Ophthalmology and Strabismus (AAPOS), the American Association of Cataract and Refractive Surgery (ASCRS), the American Society of Retina Specialists (ASRS), and Stanford University’s Byers Eye Institute.

Representatives from the FDA, various ophthalmology subspecialties, eye care industry, and health care informatics were present to share insights and to sit on panels discussing questions on the safety and effectiveness of a digital health device providing a diagnosis; the safety and effectiveness concerns for an ophthalmic digital health device used in various eye care or nonclinical environments; the most effective methods of mitigating risks for an ophthalmic digital health device; and the assets, threats, and vulnerabilities that should be considered and identified as a threat to the privacy of a patient by ophthalmic digital health device developers.

In his presentation on mobile medical devices and digital health, Zachary Bodnar, MD, Ophthalmology Innovation Fellow and Vitreoretinal Fellow at Stanford University, stressed that when developing ophthalmology apps, there are many risk profiles to consider. Risk mitigation can include restricting installation to validated configurations, robust quality assurance, and acknowledging the importance of human factors (user interface and user experience, documentation, clear error reporting). But, he said, despite the challenges, the benefits are worth the effort.

“The smartphone is ubiquitous,” he said, “It’s great for medical devices.” The smartphone brings technology to the point of care, improves efficiency and automation, streamlines communication, and provides insights with home monitoring, he added.

Additional presentations and discussion focused on methods of improving image interpretation using deep learning and artificial intelligence (AI). An important concern for deep-learning and AI algorithms in ophthalmology is that results come from a “black box” that can’t be analyzed. But, said Michael Trese, MD, vitreoretinal specialist at the William Beaumont Hospital in Royal Oak, Michigan, “The black box is opening, and the gold standard is evolving.”

“This is a deep learning revolution,” said Krishna Yeshwant, a physician, programmer, entrepreneur, and General Partner with Google Ventures. The modern reincarnation of artificial neural networks are much larger and more complex, he said, with the potential to read fundus images, perform small and large cohort analyses, and collect patient-generated data.

Linda M. Zangwill, PhD, professor of ophthalmology at the University of California San Diego, predicts that algorithms will exist in the future for retinal diseases. “We will reinvent the eye exam and allow more time for patient interaction and more seamless integration with EMRs,” she said.

Telemedicine was an important part of the day’s conversation. R.V. Paul Chan, MD, a vitreoretinal specialist from the Illinois Eye and Ear Infirmary, University of Illinois at Chicago, Illinois, discussed studies under way on telemedicine for ROP. The Imaging and Informatics in Retinopathy of Prematurity (i-ROP) and Global Education Network for Retinopathy of Prematurity (GEN-ROP) groups are developing a telecertification program to help physicians identify missed ROP diagnoses and improve their accuracy.

More home-based testing could soon be a reality, evidenced by a presentation by Quinton Oswald, CEO of Notal Vision, in which he described the company’s home-based OCT telemonitoring application, the Notal Home OCT. The system, according to Oswald, uses AI to automate analysis of home-based OCT output and identifies fluid/lesion activity. In preliminary studies, compared to reading center and independent retinal specialists, home OCT had the same reliability.

The ultimate goal regarding digital ophthalmology initiatives, as generally agreed upon by panelists and attendees, is to thoroughly assess risk and then create devices, apps, systems, and algorithms that keep data safe. Clinicians will continue to provide the context that enables accurate analysis of data. Apps must also be integrated across channels so that patients can easily use them. Physicians, industry, and regulators must train a keen eye on all of these considerations and collaborate to create successful systems that improve care. Unanswered questions remain about who owns any data collected and how to safely store data.

Presentation slides and a transcript will be available at