Week in Review: Ultra-Processed Food and Glaucoma, AI-Based Oculoplastic Surgeon Finder, Self-Administered VA Test

A weekly roundup of ophthalmic news from around the web.

A junk-food diet could open the door to glaucoma. Data from 19,225 participants in the ongoing SUN Project study, which has been following a cohort of Spanish university graduates since 1999, were analyzed to evaluate potential relationships between self-reported ultra-processed food (UPF) consumption (e.g., sodas, fried foods, snack cakes) and glaucoma risk. During an average of 12.9 years of follow-up, 230 participants developed glaucoma. In a fully adjusted model, consuming ≥4 servings of UPF/day was linked with a greater risk of developing glaucoma than consuming ≤1 serving of UPF/day (hazard ratio [HR] 1.84). Of the UPF categories, sweets were the most correlated with glaucoma risk (HR 1.52), perhaps because of the resulting elevated blood glucose levels. Nutrients

Expect bugs when using AI to find oculoplastic/oculofacial surgeons. Responses varied widely when investigators asked 3 artificial intelligence (AI) large language models—ChatGPT 3.5, Google Bard, and Microsoft Bing Balanced—for recommendations on oculoplastic surgeons, oculofacial surgeons, and eyelid lift specialists in the 20 largest US cities. Of the 672 original suggestions, 133 (19.8%) were excluded due to the provider being retired, deceased, located in a different city, not a physician, or not actually existing. Only 64% of the remaining suggestions met the stated criteria. Given this relatively high rate of inaccuracy, “it is important to continually assess whether the responses these systems provide are accurate and identify any inherent biases,” say the authors. Cureus

A self-administered visual acuity test prototype could help patients who need regular checks. Researchers in Israel have developed a home-based visual acuity (VA) test, the Hadassah Self-Visual Acuity Screener (HVAS), that would allow patients with chronic conditions or limited access to eye care to be able to regularly conduct vision self-monitoring. The prototype consists of a card containing an eye chart on one side (designed to measure a VA range not covered by the commonly used Rosenbaum Pocket Vision Screener [RPVS]) and a flowchart for how to perform the test on the other side. Two hundred seventy-five study participants with best-corrected VA >1.0 logMAR used the HVAS at home. Reported VA values were significantly correlated, in terms of reliability and accuracy, with near RPVS results and were moderately correlated with distance Snellen test results. Journal of Clinical Medicine