CIs for get_predicted.coxph() when type = "expected"

[strengejacke]

@vincentarelbundock @bwiernik for coxph, type = "expected" seems to return incorrect CIs - probably because the back-transformation is not correct. Do you know how to get CIs from the predictions and SEs, when get_predicted(mod, predict = "expected")?

From ?predict.coxph:

type the type of predicted value. Choices are (...) the expected number of events given the covariates and follow-up time ("expected") (...). The survival probability for a subject is equal to exp(-expected).

Originally posted by @strengejacke in #530 (comment)

[vincentarelbundock]

It has been yeeeears since I've worked with these kinds of models, and I'm not sure exactly what should be done here. I can look into this issue eventually, but it's likely to take a couple weeks at least.

[strengejacke]

Ok, let's leave this open and work later on it.

[bwiernik]

Yeah, it's not just a inverse link for a survival model. Need to dig in to remember