11 May 2012 07:23

Re: survival analysis simulation question

```Hi Grace,

I seem to have sent an empty draft before.  Anyway, something like
this might be an approach (untested):

require(rms)
set.seed(10)
dat <- data.frame(
age = rnorm(500, 40, 10),
race = factor(sample.int(2,500,TRUE), labels = c("a", "b")))
X <- model.matrix(~ age + race, data = dat)
b <- c(-2, .1, 3)
rates <- exp(X %*% b)
# distribution of survival time
dat\$survtime  <- 100 * rexp(500, rate = rates)
# indicator for censored/observed
dat\$cens <- dat\$survtime > 10
# new survival time values with censored
dat\$survtime  <- pmin(dat\$survtime, 10)
test <- survreg(Surv(survtime, cens) ~ age + race, data = dat)
summary(test)

Cheers,

Josh

On Thu, May 10, 2012 at 7:41 PM, Grace Ma <grace.yanfei.m <at> gmail.com> wrote:
> Hi,
> I am trying to simulate a regression on survival data under a few
> conditions:
> 1. Under different error distributions
> 2. Have the error term be dependent on the covariates
>
> But I'm not sure how to specify either conditions. I am using the Design
> package to perform the survival analysis using the survreg, bj, coxph
> functions.   Any help is greatly appreciated.
>
> This is what I have so far:
> survtime  <- 10*rexp(500)  #distribution of survival time
> cens <- ifelse(survtime > 10, 0, 1) #indicator for censored/observed
> survtime  <- pmin(survtime, 10) #new survival time values with censored
> info
> age <- rnorm(200, 40, 10)  #age variable
> race <- factor(sample(c('a','b'),500,TRUE))  #categorical variable
> test <- bj(Surv(survtime, cens) ~ rcs(age,5) + race)
>
>        [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help <at> r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> and provide commented, minimal, self-contained, reproducible code.

--

--
Joshua Wiley
Ph.D. Student, Health Psychology
Programmer Analyst II, Statistical Consulting Group
University of California, Los Angeles
https://joshuawiley.com/

______________________________________________
R-help <at> r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help