Machine Learning-Driven Early Prediction of Spontaneous Preterm Birth Subtypes from Second-Trimester Plasma Metabolomic

Objective: To identify predictive second-trimester plasma biomarkers for spontaneous preterm birth (sPTB), including preterm premature rupture of membranes (pPROM) and spontaneous preterm labor (sPL). Design: Case-control study employing non-targeted metabolomics and machine learning. Setting: Single-center study analyzing archived maternal plasma samples. at Tianjin Central Hospital of Obstetrics and Gynecology, China. Population/Sample 70 pregnant women (30 term deliveries, 20 pPROM, 20 sPL) at 14–20 weeks’ gestation. Methods: Utilizing liquid chromatography-mass spectrometry (LC-MS) for metabolomic profiling, metabolite selection is performed through LASSO regression and pathway enrichment analysis. Statistical validation involves Pearson correlations (cor(), cor.mtest()) and risk stratification modeling. Main Outcome Measures Enriched biological pathways, predictive accuracy (AUC), key metabolites, and correlations between metabolites and clinical parameters. Results: Second-trimester metabolomic signatures, particularly arachidonic acid (pPROM) and secondary metabolite (sPL) pathways, enable early sPTB risk stratification. Nine metabolites linked to inflammatory activation, oxidative stress, and placental dysfunction were identified. LASSO models achieved high predictive accuracy (AUCs: 0.984 for controls, 0.964 for pPROM, 0.995 for sPL). Creatinine and LysoPC(P-16:0) correlated positively with gestational age at blood sampling (R=0.27/0.23), while phosphatidylcholine negatively correlated with maternal age (R=-0.31). Gestational age at delivery negatively correlated with BMI (R=−0.51). High-risk stratification showed declining preterm probability with advancing gestation, contrasting with stable low-risk profiles. Conclusions: Second-trimester metabolomics combined with machine learning demonstrates robust potential for early sPTB risk prediction. Clinical translation requires multicenter validation to address cohort limitations and generalize findings.