AUTHOR=Dongling Niu , Ziwei Kang , Juanling Sun , Li Zhang , Chang Wang , Ting Lei , Hongli Liu , Yanchun Zhang 

TITLE=Universal nomogram for predicting referable diabetic retinopathy: a validated model for community and ophthalmic outpatient populations using easily accessible indicators

JOURNAL=Frontiers in Endocrinology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/endocrinology/articles/10.3389/fendo.2025.1557166

DOI=10.3389/fendo.2025.1557166

ISSN=1664-2392

ABSTRACT=PurposeThis study aimed to develop and validate a universal nomogram for predicting referable diabetic retinopathy (RDR) in type 2 diabetes mellitus (T2DM) patients, using easily accessible clinical indicators for both community and ophthalmic outpatient populations.MethodsA cross-sectional study was conducted with 1,830 T2DM patients from 14 communities in Xi’an, Shaanxi, China. Participants completed questionnaires, underwent physical exams, and ophthalmic assessments. Univariate analysis and least absolute shrinkage and selection operator (LASSO) regression identified key predictors for RDR. A nomogram was developed using multivariable logistic regression. Model performance was evaluated through area under the curve (AUC), accuracy, precision, recall, F1 score, Youden index, calibration curves, and decision curve analysis (DCA). The dataset was split into training (80%) and test (20%) sets, with external validation using 123 T2DM outpatients from Shaanxi Eye Hospital.ResultsSeven key predictors were identified: serum creatinine, urea nitrogen, urine glucose, HbA1c, urinary microalbumin, diabetes duration, and systolic blood pressure. The nomogram exhibited moderate predictive accuracy, with AUCs of 0.730 (95% CI: 0.691–0.759), 0.767 (95% CI: 0.704–0.831), and 0.723 (95% CI: 0.610–0.835) for the training, test, and external validation sets, respectively. DCA showed that using the model is beneficial for threshold probabilities between 8% and 72%, supporting its broad clinical utility.ConclusionThis nomogram, based on readily available clinical indicators, provides a reliable and scalable tool for predicting RDR risk in both community and ophthalmic settings. It offers a practical solution for early detection and personalized management of RDR, with broad applicability and clinical potential.