Desktop Survival Guide
by Graham Williams
A Cox proportional hazards model fits a baseline hazard function which is then multiplicatively modified by the input variables. (The hazard is the event of interest occurring.)
It does not model the survival time as such. We can get a prediction of survival time using the coxph method of the survfit function (to be investigated). But it is usually not possible to estimate the mean survival time from the given input variables when there is censoring.