Issue link: https://www.e-digitaleditions.com/i/1501973
26 | Summer 2023 database systems to scan through official photographs such as employee IDs or driver's licenses, searching for ptosis to increase access to care for people who need it. "Obviously, we haven't deployed any of this," says Chan. "But you can imagine using this type of noninvasive imaging that runs quietly in the background to see how that scale could potentially route people to care that they might not otherwise receive. It has the ability to impact a much, much larger population than one-on-one doctor visits." AI can replace some tedious aspects of ophthalmology as well, enabling surgeons to concentrate more on surgery. Potentially, it can also enable non- ophthalmologists to function almost like ophthalmologists in certain areas, leading to more personalized patient care. "AI is not trying to replace ophthalmologists or doctors in any way," says Yi. "But instead of having a lucky encounter with your doctor—who just happens to know how retinoblastoma presents in young children—you can actually use it to search for such anomalies." The computer can be optimized to recognize many factors. Yi describes it as "a very simple black box, but one that can do almost an infinite number of things every second. And it does them exactly the way that you tell it to. So, the computer cannot be a full clinical decision-maker, but it's something that can give physicians clinical decision support." The Theoretical Approach The concept behind the second part of Yi's lab—theoretical research—is in " " At UIC, we find ourselves in a unique position of opportunity and duty to validate medical AI models to ensure that performance does not vary among differing demographics. Darvin Yi, PhD P H O T O : D I A N E S M U T N Y