Assumes Cox PH model specification.

times and lp_thresh are arbitrarily set to 0 to prevent crashing, these should be further specified.


R6::R6Class() inheriting from MeasureSurvAUC/MeasureSurv.


All measures implemented from survAUC should be used with care, we are aware of problems in implementation that sometimes cause fatal errors in R. In future updates these measures will all be re-written and implemented directly in mlr3proba.


MeasureSurvSongTPR$new(times = 0, lp_thresh = 0, type = c("incident","cumulative"))
  • times :: vector()
    If integrate == TRUE then a vector of time-points over which to integrate the score. If integrate == FALSE then a single time point at which to return the score.

  • lp_thresh :: numeric(1)
    Determines where to threshold the linear predictor for calculating the TPR/TNR.

  • type :: character(1)
    Determines the type of score, one of: 'cumulative', 'incident'. Default 'incident'.

Meta Information

  • Type: "surv"

  • Range: \([0, 1]\)

  • Minimize: FALSE

  • Required prediction: lp


See MeasureSurv, as well as all variables passed to the constructor.


Song X, Zhou X (2008). “A semiparametric approach for the covariate specific ROC curve with survival outcome.” Statistica Sinica, 18(3), 947--65.

See also