RELATED CONTENT

NEW YORK—In keeping with its objective of providing a platform for new ideas in vision technology and eyecare, the Vision Monday Leadership Summit brought together, for the first time, four prominent eyecare researchers and practitioners for a high-level discussion about how artificial intelligence can be an important tool for disease detection and diagnosis, and then improve outcomes through early intervention.



Review of Optometry’s Paul Karpecki, OD, said AI can help address unmet needs in eyecare and beyond by improving the delivery of care to underserved populations.



Paul Karpecki, OD, FAAO, director of cornea and external disease for the Kentucky Eye Institute and chief medical editor for Review of Optometry, kicked off the session with a presentation that focused on AI’s potential to enhance diagnostic accuracy, identify disease, increase efficiency and provide better treatments across a broad range of medical conditions. “AI can actually pick up unperceived images and findings,” said Dr. Karpecki. He noted that there are now AI platforms that use retinal scans to not only identify ocular diseases, but also to detect non-ocular diseases such as Alzheimers and cardiovascular disease.

Another benefit of AI is that it can alleviate the burden on the health care system by automating routine screenings and preliminary diagnoses, Dr. Karpecki said. It can also help address unmet needs in eyecare and beyond by improving the delivery of care to underserved populations.

“Twenty-five percent of counties in the U.S. don’t have an optometrist or ophthalmologist,” Dr. Karpecki observed. He said that integrating AI with remote care can help compensate for the shortage by allowing medical expertise to be shared in key categories. “By integrating AI, we’ll be able to deliver expertise worldwide, and provide better access in underserved regions.”

AI can also simplify and expedite certain daily tasks that can lead to physician burnout, Dr. Karpecki said. “If you’re spending 60 percent of your time clicking in front of a computer with your back to the patient, I would burn out. We have to get AI to help us get to the next level of where we need to be,” he remarked.

Integrating AI into an eyecare practice has non-clinical benefits as well, Dr. Karpecki noted. “When you’re trying to sell spectacles, contact lenses or sunglasses to a patient, AI can help you understand what their likes are. It fits into everything that we do.”

He proposed that “augmented intelligence” is a more useful term than AI because “we can learn a lot by combining our knowledge and our expertise with Machine Learning and Deep Learning. An AI platform can take us to the next level. That’s where the future will be. And that’s why acceleration is going to happen in AI.”

Pearse Keane, MD, a consultant at Moorfields Eye Hospital in London and professor of artificial medical intelligence at the Institute of Ophthalmology at University College London, followed with a pre-recorded video presentation in which he emphasized how eyecare is being transformed by AI. Dr. Keane, who leads a multidisciplinary clinical research group that aims to develop and apply AI in health care, said that “ophthalmology and optometry, and eyecare in general, have been at the forefront of this coming wave of AI-enabled health care. Our key idea is that we use ophthalmology as an exemplar for other medical specialties in this regard.”

Dr. Keane summarized the pioneering research program at Moorfields, one of the largest eye hospitals in the world, which began in 2015 when Dr. Keane initiated a collaboration between Moorfields and the AI company, Google DeepMind. Since then, he and his colleagues have been investigating how AI can accurately analyze OCT scans to detect early signs of retinal disease. “This has the potential to save the sight of millions of people,” said Dr. Keane. “We’re still very much in the early stages of this journey. We have now completed the first phase of going from an idea to an algorithm, but the next phase, the longer phase, is about going from code to clinic.

Dr. Keane described how Moorfields is spearheading a pioneering program known as “oculomics,” which he described as “the idea that we can use AI as a window to systemic health.” He said researchers have shown that it’s possible to determine a person’s sex and more by analyzing a retinal photograph using an artificial intelligence technique called deep learning. “You could predict the age of the patient just from a retinal photograph,” said Dr. Keane.

“You could say whether someone smokes or not, whether they’ve got diabetes or not. You can even give a pretty good measurement of their blood pressure directly from a retinal photograph. That suggests to us that there’s a lot of biological information that is actually still locked within these high dimensional images. And maybe, with the right tools and the right techniques, we can begin to tease that to actually get new insights and develop new treatments for these patients. I think this could be a real game changer.”

Dr. Keane said Moorfields is working on developing a program in the U.K. aimed at creating a large, national database that integrates data from OCT scans with other health care data, creating a comprehensive picture of a patient’s health. “I’m really excited about the promise of world-class care, empowering ODs and other ECPs,” he said.

Next James Tsai, MD, MBA, head of the newly established Center for Ophthalmic Artificial Intelligence and Human Health at the Icahn School of Medicine at Mount Sinai, described the new research that he and his colleagues are doing in the field of oculomics, which studies the microscopic/macroscopic and molecular biomarkers in eye disease that can be associated with systemic disease.

“Oculomics is a big deal,” said Dr. Tsai. “The National Eye Institute just announced that they’re dedicating at least $5 million to research projects over the coming years in oculomics.”



(L to R) Three leading eye doctors, Paul Karpecki, OD, James Tsai, MD, and Alex Martin, OD, discussed how the future of eyecare will involve the potential for AI to enhance diagnostic accuracy, identify disease, increase efficiency and provide better treatments.



Dr. Tsai cited new research by a colleague, Dr. Ted Smith, that shows that patients with macular degeneration have SDDs (Subretinal Drusenoid Deposits) that are associated with cardiovascular disease or stroke. “These patients are three times more likely to have SDDs,” said Dr. Tsai. So SDDs may be a risk factor for underlying disease in asymptomatic patients.

“The research shows that SDDs are strongly associated with the severity of heart disease as measured by functional tests such as cardiac index and loss of choroidal and outer retinal perfusion. Severe systemic vascular disease may be a mechanism for the SDD form of AMD. This ocular finding may help guide us in the detection of patients with severe cardiovascular, and maybe renovascular diseases.”

Praising AI’s diagnostic capabilities, Dr. Tsai cited new research by other colleagues showing that large language model’s responses to questions and cases about glaucoma and retinal management were as good or even better than the specialists. “This study accentuates ChatGPT’s incredible proficiency in diagnostic accuracy and completeness compared to fellowship-trained ophthalmologists in various clinical scenarios,” he said, adding that “a seasoned clinician could do better employing ChatGPT.”

Noting that “the eye is the window to the heart,” Dr. Tsai said he believes there is “incredible potential to truly help in revolutionizing cardiac function monitoring through non-invasive ocular imaging combined with AI and mathematical modeling for early detection and follow-up of cardiac diseases.”



Dr. Pearse Keane and his colleagues have been investigating how AI can accurately analyze OCT scans to detect early signs of retinal disease.



He also said that AI can be a powerful tool to diagnose potential visually disabling eye conditions, such as central retinal artery occlusion, quickly and accurately. “AI will make us make us better, more compassionate clinicians,” he concluded.

Dr. Tsai then joined in a discussion, moderated by Dr. Karpecki, that included Alex Martin, OD, a medical director for Boston Vision in Lawrence, Mass. Asked by Dr. Karpecki to describe how AI will impact the delivery of vision care, Dr. Martin discussed his work developing an information “library” to train a vision kiosk located in a public space. The kiosk would collect diagnostic data that could then be analyzed with AI.

“A lot of the research we’re doing is about diabetic disease, glaucoma and macular degeneration, and we’re going to fine-tune that over time,” said Dr. Martin. “Long term, we’ll be trying to get the people who don’t come into our clinics every year and figuring out how to get them to come in. And once they’re referred, we’ll need to see if we agree with the AI diagnosis, or are we going to come up with a different solution?

“A patient might say, ‘I’ve been screened somewhere else, and I think I have this condition.’ As eye doctors, we might respond by saying, ‘I understand where the machine picked that up, but here’s where you actually are. Here’s what we can actually do about it. Let’s have a much more engaged conversation instead of just coming in once a year for an annual eye exam. Being able to have these conversations puts us back in the compassionate role.”

Dr. Tsai agreed, noting that patients are “yearning for more information. “They’ll take all the data they get during our office visits and search it online when they get home. Then they’ll ask for a lot more information. Our challenge is how do we utilize AI to make them feel comfortable that we understand them as a whole person.”

Dr. Karpecki pointed out that the big advantage of AI for many doctors today is to eliminate the things they don’t want to do as clinicians, such as collecting and entering data into a patient’s electronic health record. He asked, “What are some of the opportunities for AI to relieve burnout among practitioners?”

Dr. Martin, who said he sees 40 patients a day and has a three-month waiting list, replied, “Having AI is going to help us be more responsive, bring the right people in and filter the people that could wait a month, like that myope who was a minus 2 last year and the year before, and will still be a minus 2 next year. I really want to see that 30-year-old glaucoma patient who has a pressure of 35 and it’s been that way for who knows how long. Being able to readjust who we see and at what time points is going to be critical.”

Dr. Tsai and Dr. Martin were then asked what advice they would give Summit attendees about where they should focus their efforts when integrating AI into their practices and businesses. Dr. Martin recommended that attendees try as many different AI products and software options as possible to get a firsthand feel for it. “There’s a lot of opportunity that’s out there. If you can be a part of the conversation, you can help build libraries. We’re only going to be as good as our libraries are going to be.”

Dr. Tsai replied, “One of the hopes of AI is being able to reduce disparities in health care. Make sure that these technologies are easy to use. Recognize that there are differences in access to broadband and other features. And remember the person who’s in an underserved community and try to design products they can utilize.”