Machine learning model predicts liver metastasis in rectal cancer

A new radiomics model utilizing multiparametric MRI and machine learning demonstrates the potential to accurately predict metachronous liver metastasis (MLM) in rectal cancer patients. In a study involving 301 patients, the random forest model outperformed traditional methods, achieving an area under the curve (AUC) of 0.919 in training and 0.901 in validation. The model integrates independent predictors, including carcinoembryonic antigen and MRI radiomics, to enhance preoperative risk stratification for individualized patient management.