Eyeworld

SUMMER 2026 | EYEWORLD | 69 G DIGITAL MOMENTUM by Liz Hillman Editorial Co-Director About the physicians Sophia Y. Wang, MD Assistant Professor of Ophthalmology Stanford University School of Medicine Stanford, California Barbara Wirostko, MD, FARVO Adjunct Professor of Ophthalmology and Biomedical Engineering John A. Moran Eye Center University of Utah Salt Lake City, Utah W hile glaucoma as a subspecialty is not as far along as retina, for example, when it comes to de- velopment and utilization of big data and AI for diagnostic and patient care decisions, Barbara Wirostko, MD, FARVO, said there is a lot of work being done in this area to advance glaucoma care to being more proactive rather than reactive. "For instance, we are interested in learning about IOP, medication, systemic risk factors, family history, OCT/OCT-A, and visual fields. How does it all fit together and shape our decisions as well as patient outcomes? It's the totality of all this information that we're hoping AI will be able to incorporate and help us make a diagnosis or inform on the best next steps for a patient," Dr. Wirostko said. "It comes down to personalized medicine." Sophia Y. Wang, MD, who describes her- self as a glaucoma specialist, clinician scientist, and clinician informaticist, said she has been working with many different big data sources to evaluate various aspects of glaucoma, such as its epidemiology, its risk factors, and to train AI models to predict glaucoma outcomes. The datasets and sources that are avail- able now "dwarf some of the datasets that we had access to before in terms of size, number of patients, and different types of data," Dr. Wang said. In recent years, information like IOP and visual acuity metrics have become more available in big datasets, and with this type of information, Dr. Wang said analysis can better evaluate outcomes of glaucoma medical and procedural treatments. "Now we also have the ability to use this huge scale data to train AI models to predict those outcomes," Dr. Wang said. In terms of current applications of such information, Dr. Wang said she sees patients in clinic who might have rare outcomes and/ or rare diseases. Prior to the availability of big datasets and analytic systems for them, scien- tists weren't able to study such rarities on a large scale. "Now we can because of the aggre- gation of multicenter data," she said. "For example, if your center only does a few hundred of some procedure a year, there's only so much you can do to study that with your own data. But on a large scale, now we can investi- gate more answers and really see how different people respond to different surgeries," Dr. Wang explained. Dr. Wang discussed one of her studies that used multicenter data to look at the outcomes of different kinds of glaucoma procedures like trabeculectomies, tube shunts, and MIGS. The study looked at multiple outcomes including postoperative IOP, how efficacy of the procedure lasted, and the need for additional surgery. "We can now train models to predict how a patient will do after a trabeculectomy or how they will do after a tube shunt," she said, noting that in the future, as accuracy of these predictions improves, it could aid in selecting the right treatment for the right patient. Dr. Wirostko has been heavily involved in analyzing data that's been coming in with the iCare HOME2 tonometry system (Icare USA). She said there are about 4,000 patients who have used or are currently using iCare HOME2. With this data, she and colleagues are beginning to see patterns in IOP fluctuations, determine data-driven care recommendations for patients, and better understand things like the impact of sleep habits, exercise, and medication use. Data is also refining use of the iCare HOME2 system itself, informing things like how often a patient should take measurements, at which hours of the day/night, and for how long. "We still don't know what is contributing to the variation in presentation and the patho- physiology of glaucoma," Dr. Wirostko said. "We know genetics plays a role, as well as IOP and risk factors like myopia, age, and lamina cribrosa stiffness, but I think all of these things still need to be incorporated into better predic- tive models." What's being applied clinically due to learnings from the data from technology like iCare HOME2, Dr. Wirostko said, is a greater appreciation of the IOP elevations, patterns, and fluctuations that are occurring (and that previ- ously went unidentified) that are contributing to disease pathology. "All of this is helping us determine who needs treatment sooner. I think it helps us identify who's being overtreated or Advanced diagnostics, big data, and digital tools giving glaucoma care momentum continued on page 70

• Blank Page
• Blank Page (1)