XGBoost-based nomogram for predicting lymph node metastasis in endometrial carcinoma

Area of Science:

• Oncology
• Medical Informatics
• Biostatistics

Background:

• Endometrial carcinoma (EC) poses a significant health challenge, with lymph node metastasis (LNM) being a critical prognostic factor.
• Accurate identification of EC patients at high risk for LNM is essential for effective clinical decision-making and treatment planning.

Purpose of the Study:

• To develop and optimize machine learning models for predicting LNM risk in EC patients.
• To enhance the accuracy of identifying high-risk LNM patients for improved clinical management.
• To provide data-driven support for surgical decisions and individualized treatment strategies in EC.

Main Methods:

• Retrospective analysis of 541 EC cases with diverse clinical and pathological variables.
• Utilized multivariate Logistic regression to identify independent risk factors for LNM.
• Applied machine learning algorithms including LASSO, XGBoost, RandomForest, and SVM for feature selection and model construction.

Main Results:

• The XGBoost model demonstrated superior performance, achieving AUCs of 0.876 (training) and 0.832 (validation).
• Calibration curve analysis confirmed the model's consistency and applicability across various risk levels.
• Decision curve analysis indicated significant clinical utility and net benefits for the XGBoost model.

Conclusions:

• Machine learning models, particularly XGBoost, effectively predict LNM risk in EC patients using clinical and pathological data.
• The developed model offers valuable insights for clinicians in surgical decision-making and personalized treatment planning.
• This approach has the potential to improve patient outcomes through more precise risk stratification and tailored interventions.