NEWS & OPINION EWAP MARCH 2024 41 Contact information Hovanesian: drhovanesian@harvardeye.com Lin: slin@jsei.ucla.edu Wang: james@creativeventures.vc AI inroads in ophthalmology W hen ChatGPT started making headlines, all sectors of the workforce experienced renewed interest in and discussion on how AI would change their profession, medicine included. Elements of AI have already made strides into medicine, and ophthalmology specifically, for some time. Ophthalmology in many ways is ripe for AI due to it being an image-heavy field with large datasets. 1 While there’s the potential for AI in diagnostic areas, James Wang, a partner at the early-stage venture capital firm Creative Ventures, which focuses on healthcare among other areas, said this application faces more regulatory hurdles. Where he thinks ophthalmology — and healthcare as a whole—will see applications for AI are in clinical aids and front - of - the - office areas. There are already tools that use AI to function as digital scribes, for example, Mr. Wang said. He also said there are some AI systems in the OR. One company, he noted, has a system that counts all the scalpels and forceps in the surgical field (using a camera and AI software), making sure all are accounted for at the end of surgery. The line starts to get blurred with clinical aids that start making recommendations, Mr. Wang said. “I’ve seen an oncology AI that suggests drug cocktails or treatment plans that are most likely to fit a specific oncology patient. It’s not making a recommendation, it’s not prescribing anything, but it’s giving some statistical correlations based on what they’ve seen with genetics, based on what they’ve seen with medical history,” Mr. Wang said. John Hovanesian, MD, envisioned similar AI tools for ophthalmology. He was involved in creating MDbackline, which originally helped physicians automate patient communications. Its first embodiment, Dr. Hovanesian said, did not involve AI, but Alcon, which purchased it in 2022, is working that into the system. “[MDbackline] was designed to be situation specific,” he said. “If you asked the patient a question, it would respond with educational material that was appropriate. It wasn’t AI but it was algorithmic. The advantage of using a system like that and why I created it was not just so we could save time in the clinic but also to give us insights in a structured way on patient sentiment. For example, there are a lot of therapeutic areas where we don’t know well how products work. … We need to collect data on which characteristics have the best outcomes, and we need to turn that around and take those outcome characteristics and apply them to incoming patients.” Dr. Hovanesian said Alcon is in the process of integrating MDbackline into its digital workflow, which is an evolving ecosystem with AI and structured logic to help physicians with cataract surgery. Dr. Hovanesian said technology like this could be used for pharmaceuticals as well. He said when patients come in with dry eye, for example, they’re often prescribed a drug. Physicians, however, don’t routinely collect feedback on how these drugs work in the real world outside of formal studies. “This is a narrow area in healthcare where AI can help us, and it’s an important one because it’s one where it would be offering quality assurance processes that we currently don’t do,” he said. Dr. Hovanesian also gave an example of how AI could help in patient education. If a patient says their dry eye is more prevalent in the morning, for example, AI could serve up educational material about the unwanted side effects of fans blowing on the patient at night. “That’s innocuous advice, nothing that would be high risk, but helpful to patients who haven’t thought about it. That kind of thing I could see us including and supplementing our care,” he said. Another broad area is record by Liz Hillman Editorial Co-Director This article originally appeared in the December 2023 issue of EyeWorld. It has been slightly modified and appears here with permission from the ASCRS Ophthalmic Services Corp.
RkJQdWJsaXNoZXIy Njk2NTg0