Differential Predictability of Indicated and Spontaneous Preterm Birth in Nulliparous Women

Background: Preterm birth, defined as birth occurring before 37 weeks of gestation, poses a significant and enduring public health challenge, with substantial emotional and financial burdens on families and society. Preterm births are categorized into two types: indicated preterm birth, due to medical conditions like preeclampsia, and spontaneous preterm birth, which involves the natural onset of preterm labor. To identify preterm births early in pregnancy, we investigated the predictive ability of machine learning models in a nulliparous (first-time pregnancy) study cohort.<br><br>Methods: Our study analyzed the Nulliparous Pregnancy Outcomes Study: Monitoring Mothers-to-be cohort (nuMoM2b). We used a novel approach called learning with privileged information-information available during training but inaccessible during evaluation, such as adverse pregnancy outcomes (APOs), post-delivery physiological information, and maternal outcomes. We developed an enhanced model, XGBoost+, which incorporates this privileged information to improve performance.&nbsp;<br><br>Findings: We selected XGBoost as our base model. Our XGBoost+ model, utilizing privileged information, achieved an AUC of 0.72. For the subcategories, XGBoost+ had similar performance to XGBoost for spontaneous preterm birth (0.68 AUC versus 0.67 AUC), but improvements were more significant for indicated preterm birth (0.78 versus 0.74). These results demonstrate the benefits of using information not typically utilized in traditional models.&nbsp;<br><br>Interpretation: Our analysis revealed preterm birth as a multifaceted issue with different risk factors for its subcategories. We achieved moderate success in predicting indicated preterm birth (AUC 0.78), though this likely reflects the model's ability to identify already-developing hypertensive and placental conditions rather than predict their occurrence substantially in advance. While spontaneous preterm birth remains challenging to predict with clinical data alone (AUC 0.68 with minimal improvement from advanced methods), our research provides insights into the fundamental differences between preterm birth subtypes and establishes that improving sPTB prediction will require novel biomarkers capturing proximal biological processes.