基于花生四烯酸水平的早产儿视网膜病变预测模型构建与验证OA

Objective To develop and validate a predictive model for retinopathy of prematurity(ROP)that incorpo-rates gestational age(GA)and serum arachidonic acid(AA)levels at a postmenstrual age(PMA)of 32 weeks.Methods This study was a prospective observational study,enrolling 120 preterm infants with a GA of＜32 weeks who were born between January 2022 and December 2024.Baseline data including GA and birth weight were collected,and their serum AA levels were dynamically detected after birth and at a PMA of 32 weeks.Preterm infants were divided into the ROP group and the non-ROP group.Taking ROP occurrence as the outcome variable,a predictive model was constructed using multivariate Logistic regression analysis,and the discrimination,calibration and clinical utility of the model were evaluated through re-ceiver operating characteristic curve,calibration curve and decision curve analysis.Results There were 52 premature in-fants(43.3％)in the ROP group and 68 premature infants(56.7％)in the non-ROP group.Both GA and birth weight were significantly lower in the ROP group than in the non-ROP group,with statistically significant differences(all P＜0.001).The AA levels in preterm infants of the ROP group were persistently lower than those in the non-ROP group.Multivariate Logistic regression analysis revealed that GA and AA levels at PMA of 32 weeks were independent protective factors for the develop-ment of ROP.The area under the curve of the model for predicting ROP in preterm infants was 0.891(95％CI:0.832-0.950),with a sensitivity of 88.5％and a specificity of 80.9％.The calibration curve demonstrated that the model-predicted probability of ROP occurrence in preterm infants was in good agreement with the actual incidence.Decision curve analysis indicated that this predictive model provided a positive clinical net benefit.Conclusion A predictive model for ROP in preterm infants incorporating GA and serum AA levels at a PMA of 32 weeks is developed in this study.The model exhibits excellent discrimination and calibration,and provides a feasible tool to facilitate the shift from universal screening to precise detection in ROP management.