Application of a combined predictive model based on lung ultrasound score trajectory changes in deciding mechanical ventilator weaning for neonatal respiratory distress syndrome: a retrospective study

Objective Neonatal respiratory distress syndrome (NRDS) often requires mechanical ventilation, and accurate prediction of extubation timing is crucial. Methods A retrospective cohort of neonates with NRDS who underwent mechanical ventilation between January 2020 and December 2024 was included. Patients were divided into success and failure groups according to reintubation within 48 h post-extubation. A predictive model was constructed by integrating LUS trajectory changes, gestational age (GA), partial pressure of oxygen (PaO 2 ), and oxygenation index (OI), with multivariate analysis performed to evaluate predictive ability. Results The results demonstrated that LUS trajectory (LUS-high: OR = 24.099, LUS-medium: OR = 6.676,), GA (OR = 0.759), PaO 2 (OR = 0.964), and OI (OR = 1.409) were significant predictors of extubation outcomes. The nomogram incorporating these four factors exhibited an area under the curve (AUC) of 0.914. The Hosmer–Lemeshow test indicated good model fit ( p = 0.624), and the calibration curve closely approximated the ideal diagonal. Additionally, decision curve analysis revealed superior net benefit for the model. The internal validation cohort confirmed the reliability of the predictive nomogram. Conclusion Dynamic LUS assessment, combined with GA, PaO 2 , and OI, effectively predicts extubation outcomes in preterm neonates with NRDS undergoing mechanical ventilation. The model could aid in risk stratification and inform extubation decisions, though external validation is necessary prior to its routine clinical application.