| 【Abstract】Objective To establish a predictive model of disability risk in patients with diabetic retinopathy, and to verify the predictive effect of the model. Methods A total of 483 patients admitted to a grade-III eye hospital in Anhui Province from June 2022 to March 2023 were selected as study subjects. The patients were randomly divided into a training set (333 cases) and a validation set (143 cases) in a 7:3 ratio. According to disability, the training set was divided into a disabled group (111 cases) and a non-disabled group (222 cases). The disability risk prediction model was constructed by binary Logistic regression analysis. Hosmer-Lemeshow test and ROC curve were used to verify the model, and a column graph was drawn. Results The incidence of disability in diabetic retinopathy was 33.33%. DR Stage (OR=0.128), periodic annual eye examination (OR=0.358), disease perception (OR=1.062), depression (OR=1.087), health promoting behaviors (OR=0.976), symptoms and visual function (OR=0.950), marital status (OR=0.316), age (OR=2.098) Is an influential factor in the risk of disability in DR Patients (P<0.05). In Hosmer-Lemeshow test, 2 =9.563, P=0.297, the area under ROC curve was 0.845 (95%CI:0.804~0.886), the maximum Yoden index was 0.554, the sensitivity was 0.738, and the specificity was 0.851. The verification results showed that In Hosmer-Lemeshow test, 2=5.575, P=0.695, the area under ROC curve was 0.877 (95%CI:0.819~0.935), the maximum Yoden index was 0.589, the sensitivity was 0.738, the specificity was 0.851, and the accuracy was 95.80%. Conclusion This model is effective in predicting the risk of disability in diabetic retinopathy patients, and can provide evidence for the identification of diabetic retinopathy patients with high risk of disability. |
