XGBoost-based prediction of spontaneous preterm birth using maternal factors and serum markers: a retrospective cohort study

Background: Spontaneous preterm birth (sPTB) is a significant global health issue, contributing to neonatal morbidity and mortality. Existing predictive models for sPTB have shown limited accuracy, highlighting the need for improved prenatal care to identify high-risk pregnancies early. This study aimed to develop a more accurate predictive model by integrating maternal factors and serum markers using machine learning techniques. Methods: A retrospective cohort study was conducted using data from the Longgang Newborn cohort between January 1, 2020, and November 31, 2024. Women who delivered a singleton birth at 28–42 weeks’ gestation were included. We excluded cases of multiple pregnancy, iatrogenic preterm birth, cervical incompetence, and deliveries before 28 weeks. Data on maternal characteristics, pregnancy outcomes, and healthcare utilization were extracted. Predictors included maternal age, body mass index, pre-existing health conditions, socioeconomic variables, smoking status, gravidity, prior abortions, prior cesarean birth, and placental factors. We employed LASSO regression for feature selection and developed prediction models using XGBoost and logistic regression. Model performance was evaluated using sensitivity, specificity, PPV, NPV, and AUC. Results: The study included 19383 participants, with 572 (2.94%) experiencing sPTB. LASSO regression identified five significant factors: age, education, type 2 diabetes mellitus, PAPPA, and AFP. The XGBoost model showed superior performance with an AUC of 0.737 in the training dataset and 0.636 in the validation dataset, outperforming the logistic model. Conclusion: Our XGBoost machine learning based model, offers a promising approach for predicting sPTB with high accuracy. The integration of maternal factors and serum markers could enhance prenatal care by identifying high-risk pregnancies earlier, potentially reducing the incidence of sPTB and its associated complications.