Authors
Guoan Liu, Yanman Zhong, Ruimei Qin, Jiaying Wang, Jinling Xie
Published in
Medicine. Volume 105. Issue 44. Pages e51292. Oct 30, 2026.
Abstract
This study aimed to investigate the association between estimated glucose disposal rate (eGDR) and cataract surgery, and to evaluate the predictive performance of eGDR using machine learning approaches. Using a retrospective cross-sectional design, the present study analyzed data sourced from the 2001-2008 National Health and Nutrition Examination Survey, which encompassed a total of 41,658 participants. Propensity score matching was used to balance the following covariates: age, gender, race, and the poverty-to-income ratio. The analytical methods used in this study included logistic regression and restricted cubic splines to examine nonlinearity relationships. Additionally, eXtreme Gradient Boosting-SHapley Additive exPlanations (SHAP) modeling was used to evaluate the importance of the predictors. An inverse dose-dependent relationship was observed: for each unit increase in eGDR, the odds of undergoing cataract surgery decreased (odds ratio = 0.914, 95% confidence interval: 0.875-0.954, P < .001). The highest eGDR quartile showed the strongest protective effect (Q4 vs Q1: odds ratio = 0.69, P = .0087). Restricted cubic spline analysis revealed a threshold of eGDR = 5.693 mg/kg/minutes. Above this threshold, the cataract surgery rate decreased linearly. Stratified analyses revealed a significant protective effect in individuals aged ≥60 years (P = .001), and observed stronger inverse associations in high-income subgroups (P = .02). eXtreme Gradient Boosting-SHAP modeling identified eGDR as a pivotal predictor (mean SHAP = 0.008), surpassing the performance of the traditional clinical indices. Elevated eGDR levels are associated with a lower rate of cataract surgery, which is modulated by age and socioeconomic status.
PMID:
43012769
Bibliographic data and abstract were imported from PubMed on 11 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 5
- Comments 0