Prediction on a Semicure Object

predict.semicure

Description

Prediction from a fitted semiparametric cure model object.

Usage

predict(object, newdata, newtime = NULL, cumhaz = F, overall = F)

Arguments

object: a fitted semicure object.

newdata: a data frame containing the values of covariates at which the prediction is required.

newtime: times at which the survival probabilities are predicted. The default for each row of newdata is the time in the row.

cumhaz: if TRUE, cumulative hazards instead of survival probabilities are predicted.

overall: if TRUE, unconditional survival probabilities instead of survival probabilities of uncured patients are predicted.

Value

An object of class predict.semicure is returned. It is a list with a component named prediction, which contains the prediction. This component itself is a list with each component for one row of newdata. Each component contains time (at which the prediction is made), surv (predicted survival probabilities or cumulative hazard rates), median (the median survival time if the corresponding patient in this row is uncured), and curerate (the estimated cure rate).

This function is a method for the generic function predict() for class "semicure". It can be invoked by calling predict(x) for an object x of the appropriate class.

It can be printed with print.

See Also

semicure, predict, SEMICURE main page.

Examples

> z <- semicure(Surv(time, cens) ~ transplant, ~ transplant, data = goldman.data) 
> predict(z, newdata = goldman.data[c(1, 50), ], newtime = 50:54) 
Call: 
predict.semicure(object = z, newdata = goldman.data[c(1, 50), ], newtime = 50: 54) 

Predicted uncured survival probabilities and cure rates 
            1                      2 
cure rate 0.263            cure rate 0.197 
median 171                 median 98 
time surv                  time surv 
50 0.911                    50 0.839 
51 0.901                    51 0.821 
52 0.891                    52 0.804 
53 0.891                    53 0.804 
54 0.880                    54 0.787