S3 predict function for pamm objects

[adibender]

predict.pamm <- function(
  object, 
  newdata, 
  type = c("response", "mrl", "hazard", "cumulative hazard", "survival probability"), 
  se.type = c("direct", "sim", "delta"), 
  ...) {}

• should it have a times argument by default?

• return value:

  • expected survival time and mrl: vector of length nrow(newdata)
  • hazard, cumulative hazard and survival probability: ???

• alternatively, only allow response and mrl since other response types are covered in add_* functions

• what if user wants the mgcv predictions, i.e. predict.gam instead of predict.pamm?

• return types (type = ...):

  • hazard ("hazard")
  • cumulative hazard ("cumu_hazard")
  • survival probability ("surv_prob")
  • expected survival time (see also Support for estimation of expected survival time and similar measures #136) ("response")
  • mean residual lifetime ("mrl")

[iMSQ]

FYI, predictSurvProb() function (page 28, https://cran.r-project.org/web/packages/pammtools/pammtools.pdf) doesn't work with gam objects (pamm fitted model by mgcv::gam). Now it works with gaussian family implemented for glm only
(https://rdrr.io/cran/pec/src/R/predictSurvProb.R).

[adibender]

thx @iMSQ. Could you open up a separate issue with a preprex regarding predictSurvProb