NeuroLecture series: Dr. Zhong-Lin Lu, Ph.D.
Transforming functional vision testing with active learning and hierarchical Bayesian modeling
Vision loss and poor eye health present a public health crisis. In a 2016 report, a joint panel from the National Academies of Sciences, Engineering, and Medicine estimates that 90M of 142M Americans over the age of 40 have vision problems due to vision impairment, blindness, refractive error, age-related macular degeneration (AMD), cataract, diabetic retinopathy, and glaucoma. However, the tools we use to measure vision don’t really tell us what people see. Our inability to see changes in vision over time presents challenges in our healthcare system. We apply artificial intelligence to transform functional vision testing. Using active learning algorithms, we have developed a new generation of vision tests of the contrast sensitivity function (qCSF), visual acuity (qVA), iconic memory (qPR), visual sensitivity change (qCD), visual field (qVFM), and reading speed (qReading). The smart test provided by qCSF has revealed contrast sensitivity deficits at intermediate spatial frequencies caused by a variety of eye diseases that are overlooked by existing tests, demonstrating the importance of qCSF for clinical practice, clinical trials and clinical standards. With hierarchical Bayesian modeling, we further improved the precision and statistical power of qCSF and qVA tests, developed a collective endpoint that optimally combines information from both qCSF and qVA to enhance clinical trial decision making, and derived informative priors to further improve the efficiency of both tests. We are building an innovative platform that spans multimodal testing, mathematical models, and high-confidence prediction and decision-making to improve the accuracy of visual assessments, shorten clinical trials, reduce drug development costs, and bring better therapies to the market, thereby improve eye health for everyone.
Thursday, September 29, 2022 at 4:30 pm to 5:30 pm
Reynolds School of Journalism, 101 UNR, 1664 N. Virginia Street, Reno, NV
- Event Type
- Category
- Departments
- Audience
- Pricing
- Website
Recent Activity
No recent activity