Development and validation of a predictive risk model based on retinal geometry for an early assessment of diabetic retinopathy

• DOI: 10.3389/fendo.2022.1033611
• Publication Date: 2022-11-21T05:12:04Z
• Authors: Minglan Wang; Xiyuan Zhou; Dan Ning Liu; Jieru Chen; Zheng Zheng; Saiguang Ling
• Keywords: Nomogram; Lasso

This study aimed to develop and validate a risk nomogram prediction model based on the retinal geometry of diabetic retinopathy (DR) in patients with type 2 diabetes mellitus (T2DM) investigate its clinical application value.In this study, we collected data 410 T2DM Second Affiliated Hospital Chongqing Medical University between October 2020 March 2022. Firstly, were randomly divided into development cohort validation ratio 7:3. Then, modeling factors selected using least absolute shrinkage selection operator (LASSO). Subsequently, was built these identified factors. Two other models constructed only vascular traits or confirm performance advantage model. Finally, performances assessed area under receiver operating characteristic curve (AUC), calibration plot, decision analysis (DCA).Five predictive variables for DR among by LASSO regression from 33 variables, including fractal dimension, arterial tortuosity, venular caliber, duration (DM), insulin dosage (P< 0.05). A presented good discrimination an AUC 0.909 training 0.876 cohort. By comparing models, parameters proven have value could improve diagnostic sensitivity specificity when combined characteristics. The displayed high consistency predicted actual probability both cohorts. DCA demonstrated that led net benefits wide range threshold be adapted decision-making.This might facilitate stratification early detection T2DM.