"Fold" arguement in cv.grpsurv does not work

[irmakon]

Hello, While doing a grouped lasso for time-to-event outcome in a stack of multiply imputed datasets with folds predefined to be the same for multiple imputations of individual patients, I realized that "fold" argument fails to work. I think the issue is caused due to calculating the loss not in cvf.surv (which takes the subsetting according to folds into account) but rather outside which does not take into account the fold structure. Thanks! Irmak