Hi,
I just started learning R to do survival analysis using coxphfit and survfit (to
compare to neural network prediction for survival). Currently, I can generate
the Cox model using N cases and then get the estimated survival times for the
same N cases with survfit.cox <- coxph(Surv(time,delta)~X1+X2+X3)
tmp <- data.frame(X1,X2,X3)
sf <- survfit(coxfit,newdata=tmp)My question is if it's possible to
perform cross-validation with the Cox model, i.e. if I have N cases, can I fit
the Cox model to N-1 cases (i.e. generate beta coefficients) and then test the
model on the Nth case to see the estimated survival? This is what I do with
neural networks.
Thank you for your help!
~Neha
[[alternative HTML version deleted]]