Hi,
Is it possible to do round-robin validation with the Cox model, i.e. fit the Cox model using N-1 cases and then test the Nth case on the trained model to see what the estimated survival would be for the unknown case?
I've tried Matlab (coxphfit) and R (coxphfit, survfit) but maybe I'm not fuly understanding the functions because it doesn't seem to work. Or is round-robin validation only possible with logistic regression (using the glmfit/glmval functions in Matlab)?
Thanks!