The RetiAGE algorithm demonstrates potential in predicting osteoporosis risk through retinal assessments, as evidenced by its application in two extensive cohorts.
This algorithm estimates retinal biological age and produces a probability-like score, allowing for both cross-sectional and longitudinal analyses. The study incorporated multivariable models that adjusted for established osteoporosis risk factors, including age, sex, calcium intake, diabetes, hypertension, smoking habits, physical activity, and glucocorticoid usage.
While the effect sizes observed are modest—indicated by a per-standard deviation hazard ratio of 1.12—the data suggests a more pronounced risk for individuals in the highest quartile, with a hazard ratio of 1.40 compared to those in the lowest quartile. The authors noted a slight improvement in discrimination, with a C-index increase of 0.050, although this remains insufficient for standalone clinical applications.
This research highlights the emerging role of retinal assessments in predicting osteoporosis risk, suggesting that while the findings are promising, further validation is necessary before integrating such algorithms into clinical practice.