R help - Jun 2006 - survest function and counting process style input

Hi,

I have fitted a cox model with time-varying covariates (counting process style)
using the cph function of the Design package.
Now I want to know the survival probability of a single individual at every time
of his history.

I know the survest function, but I am not sure how to interpretet its output
when using time-varying covariates.
When the input consists of multiple values, does survest just compute the
probabilities as if it are independent individuals or can/does it take in
consideration that it is the history of a single individual?
Is the latter even possible with a counting process cox model?

An example: Individual x has a history of 3 months and the cox model is fitted
with two time-varying covariates: a & b

testcase <- data.frame(a =c(4,5,2), b = c(1,0,1))
survest(coxmodel, testcase, time = c(1,2,3))

Is this the right way to compute the probabilities of a single individual?

Thank you in advance!

Regards, 

Laurens Alberts


 

	[[alternative HTML version deleted]]