riskRegression (version 2023.12.21)

saveCoxConfidential: Save confidential Cox objects

Description

Save confidential Cox objects

Usage

saveCoxConfidential(object, times)

Arguments

object

An object of class coxph.

times

The times at which we want to predict risk.

Details

This function can save coxph objects such that we do not need to export the data on which it was fitted at given times.

Examples

library(survival)
library(lava)
set.seed(18)
trainSurv <- sampleData(300,outcome="survival")
testSurv <- sampleData(40,outcome="survival")
fit = coxph(Surv(time,event)~X1+X2+X3+X7+X9,data=trainSurv, y=TRUE, x = TRUE)
u=saveCoxConfidential(fit,times=3)
if (FALSE) {
# write object as plain text file
sink("~/tmp/u.R")
cat("U <- ")
dput(u)
sink(NULL)
# reload object
source("~/tmp/u.R")
class(u) <- "CoxConfidential"
}
predictRisk(u,newdata=testSurv)
cox1 = coxph(Surv(time,event)~strata(X1)+X2+X3+X7+X9,data=trainSurv, y=TRUE, x = TRUE)
z<-saveCoxConfidential(cox1,c(2,5))

dput(z) ## get output to copy object

all.equal(predictRisk(z,newdata=testSurv),
          predictRisk(cox1,newdata=testSurv,times=c(2,5)))

cox2 = coxph(Surv(time,event)~X1+X2+X7+X9,data=trainSurv, y=TRUE, x = TRUE)
z<-saveCoxConfidential(cox2,c(2,5))

all.equal(predictRisk(z,testSurv),predictRisk(z,testSurv,c(2,5)))