st: Cross-validation of survival analysis models

[Kim Peeters <kimpeeters84@yahoo.com>]

Dear Stata users,

I have a question regarding the validation of survival
analysis models. I would like to cross-validate the specified survival analysis
model.


The dataset that I am using is fairly large and included
data from 1988m1 to 2008m12. All information was measured once a month and as such
the covariates are time-varying. Two types of failures may occur, either event
1 or event 2. Subjects could suffer from multiple failures and the occurrence of
one event does not exclude the potential occurrence of the other event. Furthermore,
subjects were at risk during the entire period. All subjects that were included
in the dataset at the onset of the period suffer from left censoring and all
subjects that were in the dataset at 2008m12 were right censored. During the
observational period, subjects could be added or removed, the former causing
left-censoring and the latter right-censoring.

Because I am not yet very familiar with survival analysis,
any ideas or help on how to cross-validate survival analysis models will be highly
appreciated.

Thank you.

Kim Peeters
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/