Development and Validation of a Machine Learning-Based Clinical Model for Predicting Rupture in Ectopic Pregnancy: A Web-Based Nomogram Approach.

Objective: The aim of this study is to develop a predictive model for rupture-associated bleeding in ectopic pregnancy (EP) and to construct a web-based nomogram to support early clinical intervention in women at elevated risk. Methods: Clinical data were retrospectively collected from 543 women with EP at Hexian Memorial Affiliated Hospital of Southern Medical University, Guangzhou, China, between June 2019 and June 2022. Among these, 58 cases were confirmed intraoperatively to have experienced rupture with bleeding. The cohort was randomly divided into training (70%) and validation (30%) subsets. Key predictive variables were selected using the Extreme Gradient Boosting (XGBoost) algorithm, guided by SHapley Additive exPlanations (SHAP) values. Model performance was assessed using the area under the receiver operating characteristic (ROC) curve, calibration analysis, decision curve analysis (DCA), and clinical impact curve (CIC). A web-based nomogram was subsequently developed for clinical implementation. Results: Seven predictive variables were identified and used to construct the model. The ROC curve yielded an area under the curve (AUC) of 0.941 (95% CI: 0.882-0.968) in the training subset and 0.970 (95% CI: 0.9405-0.990) in the validation subset. Calibration curves demonstrated strong concordance between predicted probabilities and observed outcomes. DCA indicated a clinically meaningful predictive probability range between 1% and 94.82%. A dynamic, web-based nomogram was created to facilitate practical application. Conclusion: A clinically applicable predictive model for rupture in EP was developed and validated, incorporating seven key variables. The web-based nomogram enables early risk stratification and intervention, potentially reducing the incidence of rupture-related complications.