Development and Evaluation of a Predictive Tool for Second-Pregnancy Preterm Birth and Cervical Insufficiency in Patients With a History of Cervical Insufficiency

Background: Patients with cervical insufficiency are at a higher risk of preterm birth or recurrent cervical dysfunction in a subsequent pregnancy. Exploring key risk factors and developing predictive models is of great clinical significance. Methods: Patients diagnosed with cervical insufficiency during their first pregnancy were included and categorized according to whether they experienced preterm birth during their second pregnancy. Lasso regression was used for feature selection, followed by logistic regression (LR) and eXtreme Gradient Boosting (XGBoost) models to analyze the factors influencing the two outcomes. Model performance was evaluated through metrics such as area under the curve (AUC), accuracy, and recall rate. Model interpretation was performed using SHapley Additive exPlanations (SHAP) values, and a risk scoring model was constructed based on the core features to evaluate predictive performance. Results: The XGBoost model demonstrated the best performance in predicting preterm birth (AUC = 0.745), whereas the LR model achieved a higher AUC (0.777) for predicting cervical insufficiency. The risk model based on core features demonstrated an AUC of 0.711 for predicting preterm birth and an AUC of 0.686 for predicting cervical insufficiency. Conclusions: Cervical length at 16–18 weeks and 22–24 weeks, along with preterm birth and cervical laceration during first pregnancy, are key predictive factors of second-pregnancy preterm birth and cervical insufficiency. The risk model derived from multi-model analysis can effectively identify high-risk pregnancies.