A principled machine learning framework improves accuracy of stage II colorectal cancer prognosis

Average AUROC and standard deviation (for n？=？200) of trained classifiers on the training set using 20-times repeated tenfold cross-validatio