Integrating epigenetic and metabolic indicators for non-invasive endometrial cancer triage: a machine learning approach

The authors aimed to develop a non-invasive machine learning model that integrates epigenetic and metabolic indicators for the triage of endometrial cancer in patients with abnormal uterine bleeding. They identified CDO1 methylation as a significant biomarker and created a support vector machine model that demonstrated promising diagnostic performance, achieving an area under the curve of 0.904 in validation. This model facilitates personalized risk stratification and clinical decision-making, although external validation is needed for absolute risk estimation.