Answers are grounded in this paper and cite their sources.
Demonstrates overall survival prediction in vulvar cancer, indicating potential for personalized treatment planning.
e17651 Background: Vulvar cancer is a rare gynecologic malignancy, and accurate prognostic tools for avoiding under- or over-surveillance remain limited. We developed and externally validated an explainable machine learning model to predict overall survival (OS) in patients with vulvar cancer. Methods: We identified 12,764 patients from the SEER database (2004–2021) and an external Chinese multicenter cohort (n = 361; 2003–2023). Candidate predictors were selected in the training set using univariable and multivariable Cox regression. Five predictive models—Cox proportional hazards model (CoxPH), random survival forest (RSF), extreme gradient boosting (XGBoost), support vector machine (SVM), and Light Gradient Boosting Machine (LightGBM)—were developed using the training set. Model performance was evaluated using the C-index, time-dependent AUROC at 1, 3, and 5 years, Brier score, calibration curves, and decision curve analysis (DCA). Model interpretability was assessed using SHapley Additive exPlanations (SHAP) to provide transparent, patient-level explanations of key drivers of predicted risk. Results: Among the five models, RSF achieved the best overall performance, with a C‐index of 0.7502 in the internal validation set and 0.7500 in the external validation set. The 1‐, 3‐, and 5‐year AUROCs were 0.830, 0.817, and 0.810 in the internal validation set, and 0.770, 0.813, and 0.787 in the external validation set. SHAP interpretation identified age, FIGO stage, and lymph node involvement as the leading contributors to OS prediction. Conclusions: We developed and externally validated an explainable RSF-based OS prediction model for vulvar cancer using a large population-based dataset and a multicenter Chinese cohort. The model showed robust discrimination, calibration, and potential clinical utility, supporting individualized risk stratification and follow-up planning.
No takes yet. Share an insight, caveat, or question.
Shi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: