How did these researchers determine the confidence interval of the AUROC using resampling but without retraining the model?

Open in new window