To validate the Graves ophthalmopathy quality of life (GO-QOL) questionnaire in screening DON and to construct an effective model. A total of 194 GO patients were recruited and divided into DON and non-DON (mild and moderate-to-severe) groups. Eye examinations were performed, and quality of life was assessed by the GO-QOL questionnaire. The random forest, decision tree model, receiver operator characteristic (ROC) curve, accuracy and Brier score were determined by R software. In GO-QOL, age, best corrected visual acuity (BCVA), exophthalmos, CAS, severity, and Gorman score were found to be factors related to visual function scores. On the appearance scale, gender, duration of GO, BCVA, exophthalmos, CAS and severity of GO were relevant. Both the visual function scores and appearance scores were significantly lower in DON groups than in non-DON groups (33.18 ± 24.54 versus 81.26 ± 17.39, 60.08 ± 24.82 versus 76.14 ± 27.56). The sensitivity, specificity, and AUC of the visual function scores were 91.1%, 81.7% and 0.939, respectively Visual function scores were used to construct a decision tree model. The sensitivity, specificity, and AUC of the model were 92.9%, 88.0% and 0.941, respectively, with an accuracy of 89.7% and a Brier score of 0.024. Visual function scores were qualified as a screening method for DON, with a cutoff point of 58. A multifactorial screening model based on visual function scores was constructed.
Read full abstract