Retina specialists map AI's clinical gains and regulatory gaps

Artificial intelligence is accelerating how retina specialists analyze imaging, design clinical trials, and screen for eye disease, but the technology faces a multi-year regulatory bottleneck before it can deliver personalized treatment insights at the point of care, according to ophthalmologists and vision researchers interviewed by Ophthalmology Times.

Why it matters

AI tools are already reshaping research workflows and imaging interpretation in retina care, yet the gap between what works in the lab and what clinicians can legally deploy remains wide. The delay has practical consequences: physicians want to show patients individualized disease trajectories and treatment responses, but no FDA-cleared tool exists to do so. Meanwhile, questions around data ownership, equitable access, and patient self-diagnosis remain unresolved.

Clinical trials get new endpoints

Hasenin Al-khersan, a retina specialist with Retina Group of Florida, said AI is enabling trials to stratify patients based on photoreceptor loss rather than older markers such as geographic atrophy or fundus autofluorescence. Because photoreceptor loss precedes the atrophy that develops later, trials can now catch disease progression earlier, when intervention still matters. Companies including Biogen and Apellis entered the retina space relying on what Al-khersan called "almost archaic" measurement tools; AI-driven endpoints represent a material shift.

Arshad M. Khanani, director of clinical research at Sierra Eye Associates, said AI can also identify trial candidates through imaging biomarkers and electronic health record data, which speeds recruitment and reduces screen failure rates.

Imaging analysis and global screening

Joel Pearlman of Retina Consultants Medical Group said automated, consistent analysis of retinal imaging helps clinicians quantify treatment efficacy. Mina Massaro-Giordano, a professor of ophthalmology at NYU Langone Health, noted that AI can surface patterns across large image sets that humans miss, such as changes in tear film evaporation on the ocular surface, though she cautioned that results depend on accurate, well-curated training data.

Paul Nderitu, a retinal specialist at Moorfields Eye Hospital, argued that AI's largest global impact may come from lowering the cost and skill threshold needed to detect cataracts, refractive errors, glaucoma, diabetic eye disease, and macular degeneration using handheld or low-cost devices. AI serves as "the gateway to detecting them" before referral, Nderitu said, expanding access in settings that lack specialists.

Khanani emphasized that AI-assisted screening flags patients who need evaluation but does not replace physician judgment. "It's not like AI is going to completely take over the field of medicine—we're always going to need human doctors working with it," he said.

Regulatory and operational barriers

Al-khersan identified the regulatory pathway as the biggest obstacle to closing the gap between research capability and clinical deployment. A device or pharmaceutical company—or the field itself—must invest the time and money required to gain FDA approval for a clinic-ready tool, a shift Al-khersan estimated is "probably at least a few years out." He cited personalized treatment trajectory visualization as a capability colleagues want but cannot yet access.

Andrew Pucker, chief development officer at contract research organization Mentra Health, described a more operational use case: AI agents built with partner Tilda Research now handle tasks such as categorizing trial documents and flagging errors in an electronic trial master file. The company is expanding that approach to onboarding and electronic data capture.

Privacy, equity, and patient self-diagnosis

Khanani raised confidentiality as a central concern, including who owns patient data, where it is stored, who has access, and how it might be used beyond its original purpose.

Nderitu warned that unequal access to AI—particularly as tools move to paid subscription models—could widen care disparities if the field does not ensure broad representation during development. He called the dynamic "AI poverty" and argued that AI should be available "at different rungs of affordability."

Khanani said patients using tools such as ChatGPT to self-diagnose or self-treat is "a really bad idea," because the algorithms draw from both reliable and unreliable sources and can produce wrong answers. Nimesh Patel of Mass Eye and Ear offered a different view, saying he encourages patients to research their conditions using AI before visits, arguing that baseline knowledge makes treatment conversations easier.

Pearlman framed AI as a collaborator rather than a replacement: "AI will not replace doctors; it'll only replace doctors who don't use AI."

Ophthalmology Times first reported these perspectives in a collection of interviews with retina specialists and vision researchers.