Predictive Effect of Nomogram on Diabetes Retinopathy in Patients with Diabetes

YAO Xinxia; ZHANG Xinjie; GUO Jiajia; DUAN Yang; LIU Hongzhou; YAN Shuangtong

doi:10.3969/j.issn.2095-3097.2023.06.008

Abstract：Objective To investigate the prediction of diabetic retinopathy(DR)by nomogram model.Methods In this study,503 adult diabetic patients who met the inclusion and exclusion criteria for fundus photography in a hospital be-tween 2017 and 2018 were finally retrieved from the dryad website.Univariate and multivariate Logistics regression analysis models were used to analyze the risk factors of DR In diabetic patients.Based on the analysis results and clinical experience,a histogram model for predicting the risk of DR In diabetic patients was constructed.Receiver operating characteristic(ROC)curve and calibration curve were used to evaluate the prediction effect of the nomogram.Results Patients with and without DR Diabetes gender,duration of diabetes,combined hypertension,systolic blood pressure,high density lipoprotein cholester-ol,low density lipoprotein cholesterol(LDL-C),alanine aminotransferase,aspartate aminotransferase,albumin,urea,creati-nine,urinary protein,urinary albumin,urine albumin-to-creatinine ratio(UACR),glomerular filtration rate,the difference of chronic kidney disease grade was statistically significant(P<0.05,P<0.01).In this study,gender,course of diabetes,combined hypertension,glycosylated hemoglobin,LDL-C and UACR were included in the column graph model.The area un-der the ROC curve of the nomogram model to predict the risk of DR In diabetic patients was 0.724(95%CI 0.676,0.772),and the calibration curve was close to a line with slope of 1.Conclusion The nomogram model can be used to predict the risk of DR In diabetic patients,and can identify high-risk groups early,so that timely intervention can improve the prognosis of patients.