I have a data frame including various continuous variables (size of a lymph node measured in different ways) and an outcome (disease-free interval).

I have used the intcensROC package to assess the suitability of the size of the lymph node as a marker of disease-free interval with interval censored data.

U <- ASAC_tot$Left # left side of the interval censor data
V <- ASAC_tot$Right # right side of the interval censor data
Marker <- ASAC_tot$lymph.node.size  # lymph node size
Delta  <- ASAC_tot$CENSOR_DFI # censuring status
pTime <- 180 # to determine to predicted time
res <- intcensROC (U, V, Marker,  Delta, pTime, gridNumber = 500)
auc <- intcensAUC(res)

when I run this function, I end-up with the same AUC whatever the value of pTime...I find it strange...does anyone have any idea why ?

Also, even if I doubt there is a simple solution, is-there any way to determine a cut-off for the best lymph node size ? (no worries, the determination of the cut-off is ONLY to destroy its use :))


  • $\begingroup$ Hi, welcome to CV.SE! Since this is a programming question, perhaps you would benefit from posting this to Stack Overflow as well. $\endgroup$ – LmnICE Sep 5 '20 at 19:53
  • $\begingroup$ A key question is whether it is an all-time c-index (AUROC) vs. a time-specific c-index. Also note that the c-index (concordance probability) is not sensitive enough for comparing two models: fharrell.com/post/addvalue $\endgroup$ – Frank Harrell Sep 5 '20 at 20:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.