Maternal and Neonatal Determinants of Respiratory Outcome Following Second-Trimester PPROM: A Multi-Domain Machine...

Area of Science:

• Perinatology and Neonatology
• Computational Biology and Machine Learning
• Reproductive Medicine

Background:

• Preterm premature rupture of membranes (PPROM) before 32 weeks gestation with prolonged latency leads to significant neonatal morbidity, including Dry Lung Syndrome (DLS), pulmonary hypoplasia (PH), bronchopulmonary dysplasia (BPD), and death.
• Current statistical methods inadequately capture the complex interplay of amniotic fluid dynamics, inflammation, and microbial load for accurate risk stratification.

Purpose of the Study:

• To develop and validate machine learning models for predicting neonatal outcomes in pregnancies with second-trimester PPROM and prolonged latency.
• To identify distinct, multi-domain risk architectures for specific neonatal morbidities (DLS, PH, BPD, intraventricular hemorrhage (IVH), neonatal death) that are not apparent with traditional statistical approaches.

Main Methods:

• Retrospective analysis of 66 pregnancies with second-trimester PPROM and latency >14 days.
• Elastic Net and Random Forest models trained on six predictor domains (gestational age, latency, amniotic fluid, inflammation, vaginal microbiome, postnatal factors).
• Models were evaluated separately for antenatal (Model A) and combined antenatal/postnatal (Model B) predictors.

Main Results:

• Pulmonary hypoplasia risk was linked to the persistence and timing of oligohydramnios (SDP < 1 cm).
• Antenatal model (Model A) for Dry Lung Syndrome achieved AUC 0.776, driven by gestational maturity and inflammatory status.
• Intraventricular hemorrhage showed high predictability (accuracy 0.863) with amniotic fluid dynamics and microbiological burden as key predictors.

Conclusions:

• Machine learning uncovers outcome-specific risk profiles following PPROM, invisible to conventional methods.
• Longitudinal amniotic fluid trajectory is a primary antenatal predictor of structural pulmonary morbidity.
• Microbiological burden independently influences neurological risk, supporting the development of integrated ML-based risk stratification tools for antenatal counseling.