Preterm infants are at high risk for neonatal acute respiratory distress syndrome (ARDS) due to physiological immaturity and multiple factors. This study aimed to develop and validate a prediction model for this population. This study retrospectively analyzed clinical data from preterm infants aged 28 ~ 37 weeks who required mechanical ventilation (MV) within 7 days after birth. These infants were admitted to 2 hospital Neonatal Intensive Care Units (NICU). Clinical data, including blood parameters and arterial blood gas indicators, were collected. Least Absolute Shrinkage and Selection Operator (LASSO) and multivariate logistic regression identified independent predictors to develop a nomogram model. Model performance was evaluated using ROC curves, calibration, and decision curve analysis (DCA), with external validation. The results indicated that maternal education level, gestational age, birth weight, SIRI, and arterial PaCO₂ within 1 h after admission were independent predictors for ARDS diagnosis in preterm infants. Integrating these variables into the prediction model yielded an AUC of 0.890 in the training set and 0.845 in the validation set, with a specificity of 0.816. Within the predicted probability range of 0.05–0.95, the model demonstrated superior predictive performance compared with any individual predictor alone. Conclusion: Based on prenatal risk factors and early postnatal blood gas and biochemical indicators, this study developed a novel risk prediction model for ARDS in moderate-to-late preterm infants. It provides a reference for early identification and precise intervention of ARDS in this population.<Table Float="No" ID="Taba"> <tgroup cols="1"> <colspec align="left" colname="c1" colnum="1" /> <tbody> <row> <entry align="left" colname="c1"> What is Known: • Existing predictive models for ARDS are largely based on term or late-preterm infants. Preterm infants at 28 ~ 37 weeks of gestation, however, present a diagnostic challenge owing to lung immaturity and often atypical symptoms. There are currently no validated tools for specifically assessing ARDS risk in this population. </entry> </row> <row> <entry align="left" colname="c1"> What is New: • We developed an interpretable nomogram incorporating preterm gestational age, birth weight, SIRI, arterial PaCO₂, and maternal education level to provide individualized ARDS risk assessment for moderate-to-late preterm infants. By visualizing each predictor’s contribution at the bedside, it offers clinical transparency and a unique advantage over prior generalized models, filling a critical gap in tools for this vulnerable population. </entry> </row> </tbody> </tgroup> </Table> Graphical Abstract

Development and validation of a predictive model for acute respiratory distress syndrome in moderate-to-late preterm infants: a multicenter retrospective study

期刊论文
发表日期：

2026年4月29日

查看全文

Chao Zhang,
Wenxiao Xiong,
Kangjin Diao,
Zhixing Gao,
Wenlong Zhang,
Jiajia Guo,
Tian Yao,
Yong Ji,
Huaiqing Yin,
Shirun Wu

摘要

Preterm infants are at high risk for neonatal acute respiratory distress syndrome (ARDS) due to physiological immaturity and multiple factors. This study aimed to develop and validate a prediction model for this population. This study retrospectively analyzed clinical data from preterm infants aged 28 ~ 37 weeks who required mechanical ventilation (MV) within 7 days after birth. These infants were admitted to 2 hospital Neonatal Intensive Care Units (NICU). Clinical data, including blood parameters and arterial blood gas indicators, were collected. Least Absolute Shrinkage and Selection Operator (LASSO) and multivariate logistic regression identified independent predictors to develop a nomogram model. Model performance was evaluated using ROC curves, calibration, and decision curve analysis (DCA), with external validation. The results indicated that maternal education level, gestational age, birth weight, SIRI, and arterial PaCO₂ within 1 h after admission were independent predictors for ARDS diagnosis in preterm infants. Integrating these variables into the prediction model yielded an AUC of 0.890 in the training set and 0.845 in the validation set, with a specificity of 0.816. Within the predicted probability range of 0.05–0.95, the model demonstrated superior predictive performance compared with any individual predictor alone. Conclusion: Based on prenatal risk factors and early postnatal blood gas and biochemical indicators, this study developed a novel risk prediction model for ARDS in moderate-to-late preterm infants. It provides a reference for early identification and precise intervention of ARDS in this population.

What is Known:

• Existing predictive models for ARDS are largely based on term or late-preterm infants. Preterm infants at 28 ~ 37 weeks of gestation, however, present a diagnostic challenge owing to lung immaturity and often atypical symptoms. There are currently no validated tools for specifically assessing ARDS risk in this population.

What is New:

• We developed an interpretable nomogram incorporating preterm gestational age, birth weight, SIRI, arterial PaCO₂, and maternal education level to provide individualized ARDS risk assessment for moderate-to-late preterm infants. By visualizing each predictor’s contribution at the bedside, it offers clinical transparency and a unique advantage over prior generalized models, filling a critical gap in tools for this vulnerable population.

Graphical Abstract