# Forecasting survival probability using Cox Regression

I'm able to obtain predicted survival probabilities of cox regression using either survfit.coxph or predictSurvProb from pec package. However, using these approaches I'm unable to predict the probability of new observation which has the time set after the study, or the end time of dataset used to construct the model, e.g. the period of study is from 0 to 1000, and the time for new observation is 1200.

Is this the limitation of cox regression? Otherwise, is there any alternative to calculate the aforesaid probability?