Desktop Survival Guide
by Graham Williams
We might want to predict the time until event (e.g., death) for new observations. The function survreg can do this. A positive coefficient for an input variable means a longer survival time, which is usually a good outcome.
Or we might want to predict the death rate. The function coxph can do this. A positive coefficient for an input variable means a higher death rate, which is usually a bad outcome.
The different interpretations of the sign of the coefficient for the survreg model and the coxph model leads to some confusion. In the former, a positive coefficient is a good outcome and in the latter it is a bad outcome.
For either type of model the predict function can be used to obtain predictions.