Deciphering the predictors of endometrial nonbenign lesions in asymptomatic postmenopausal women via explainable machine learning

Key Points

• Endometrial nonbenign lesions were accurately predicted using a machine learning model, improving diagnosis accuracy.
• The nomogram presented significant insights, aiding clinicians in personalized treatment plans and avoiding unnecessary surgeries.
• We utilized an LR-based nomogram model with SHAP for interpretability to analyze predictors of lesions in women.
• This modeling may enhance clinical decision-making, highlighting the need for further validation in diverse settings.

Abstract