A higher eGDR, a marker of insulin sensitivity, may indicate lower risk of DR. The study, published in the Journal of Diabetes & Metabolic Disorders, found a strong inverse association between eGDR and DR, suggesting its clinical potential for early risk stratification.

The analysis included 954 adults with diabetes from the 2015–2020 National Health and Nutrition Examination Survey. eGDR values were evaluated using regression and machine learning methods to explore their relationship with retinal outcomes.

Each one-unit increase in eGDR was linked to a 21% reduction in DR risk (OR = 0.79; p < 0.001). Quartile-based analysis revealed the strongest protection in the third quartile (OR = 0.53; p = 0.020), with a significant linear trend (p < 0.001). Restricted cubic spline modeling confirmed a non-linear association (p = 0.029) with an inflection point at eGDR = 3.92, beyond which the protective effect plateaued.

Among eight predictive algorithms tested, XGBoost achieved the highest discrimination (AUC = 0.894). Feature importance analyses ranked eGDR among the strongest predictors of DR, reinforcing its role as a practical biomarker.

These findings highlight eGDR as a feasible tool for identifying patients at lower risk of diabetic retinopathy and optimizing early screening and intervention strategies.