Displaying 1 result from an estimated 1 matches for "survpr".
Did you mean:
survr
2018 May 24
1
Predictions from a Cox model - understanding centering of binary/categorical variables
...,data=ovarian)
If I want to make predict survival for a new set of individuals at 100 days then that is trivial using predict.coxph e.g.:
newdata <- data.frame(futime=rep(100,5),fustat=rep(1,5),age=c(45,50,55,60,65),rx=c(1,2,1,2,1))
preds <- predict(ova_mod,newdata,type="expected")
survpreds <- exp(-1*preds)
survpreds
[1] 0.9967635 0.9969739 0.9859543 0.9868629 0.9401437
However, due to centering I believe, I am finding this a bit difficult to replicate by hand.
To replicate the analysis I need the baseline survival at 100 days, the regression coefficients, and the mean/proport...