SimSurv(10)
SurvData=SimSurv(100,cens.baseline=1/10,surv.baseline=2)
Hist(SurvData$time,SurvData$status)
prodlim(Hist(time,status)~1,data=SurvData)
plot(prodlim(Hist(time,status)~1,data=SurvData))
SurvData=SimSurv(100,cens.baseline=1/10,surv.baseline=2,surv.coef=c(-1,-2),
cova=list( X.exp = list(dist = "rexp", rate = 0.4),
X.bernoulli = list(dist = "rbinom", size = 1, prob = 0.3)))
SurvData=SimSurv(100,cens.baseline=1/10,surv.baseline=2,surv.coef=c(-1,-2),cens.coef=c(0,1),
cova=list( X.exp = list(dist = "rexp", rate = 0.4),
X.bernoulli = list(dist = "rbinom", size = 1, prob = 0.3)))
set.seed(8)
SurvFrame <- SimSurv(N=100,
surv=list(dist = "rweibull",args = list(shape=1),baseline = 1/100,link="exp",coef=1),
cens=list(dist = "rexp",args=NULL,baseline=1/1000,link="exp",coef=0,max = 150,type = "interval",lateness=1,unit=10),
cova=list(Age = list("rnorm", mean = 0, sd = 2),Sex = list("rbinom", size=1,prob=.5)),
keep.uncensored=TRUE,
method=c("simulation"),
verbose=1)
library(survival)
coxph(Surv(time,status==0)~X.exp+X.bernoulli,data=SurvData)
coxph(Surv(time,status)~X.exp+X.bernoulli,data=SurvData)
# Simulate survival times based on a parametric survival fit
library(survival)
data(ovarian)
fitSurv <- survreg(Surv(futime,fustat)~age+rx,data=ovarian)
fitCens <- survreg(Surv(futime,1-fustat)~age+rx,data=ovarian)
# note: survreg fits the accelerated failure time model version of the
# Weibull survival regression model.
# The parameters need to be transformed in order to
# simulate from a fitted model using SimSurv:
survCoef <- -fitSurv$coef[-1]/fitSurv$scale
survBaseline <- exp(-fitSurv$coef["(Intercept)"]/fitSurv$scale)
survShape <- 1/fitSurv$scale
censCoef <- -fitCens$coef[-1]/fitCens$scale
censBaseline <- exp(-fitCens$coef["(Intercept)"]/fitCens$scale)
censShape <- 1/fitCens$scale
X <- model.frame(~age+rx,data=ovarian)
simuData <- SimSurv(N=26,
cova=X,
surv.coef=survCoef,
surv.baseline=survBaseline,
surv.shape=survShape,
cens.model="Cox-Weibull",
cens.coef=censCoef,
cens.baseline=censBaseline,
cens.shape=censShape)
simuFit <- survreg(Surv(time,status)~age+rx,data=simuData)
cbind(coef(simuFit),coef(fit))
plot(prodlim(Surv(time,status)~rx,data=simuData),confint=F)
plot(prodlim(Surv(futime,fustat)~rx,data=ovarian),confint=F,add=TRUE,lty=2)
## another example
surv.coef=c(0.1, 0.03, -0.5, 0.5)
cens.coef=c(0.1, 0.03,-0.5, 0.5)
variables=list(
x1=list(dist="rnorm", mean=26, sd=2),
x2=list(dist="rnorm", mean=50, sd=7),
x3=list(dist="rbinom", size=1, prob=0.2),
x4=list(dist="rbinom", size=1, prob=0.5))
surv.base<-0.00001
cens.base<-0.00009
sim.dat<-SimSurv(N=500,
cova=variables,
surv.coef=surv.coef,
cens.coef=cens.coef,
surv.model="Cox-exponential",
cens.model="Cox-exponential",
surv.baseline=surv.base,
cens.baseline=cens.base,
verbose=F)
## adjusting the censoring percentage
## for a given survival setting
SNP1 <- rbinom(181,2,.3)
stage <- rbinom(181,2,.5)
setup <- list(N=181,surv.baseline=1/100, cova=list(a=SNP1,b=stage), surv.coef=c(1.3,1.8), cens.baseline=1/100)
d <- do.call("SimSurv",setup)
set30 <- prodlim:::find.baseline(x=.3,setup)
d30 <- do.call("SimSurv",set30)
set70 <- prodlim:::find.baseline(x=.7,setup)
d70 <- do.call("SimSurv",set70)
## to visulize the shape of the censoring
## distribution do this:
plot(prodlim(Hist(time,status)~1,data=d30,rev=TRUE))
plot(prodlim(Hist(time,status)~1,data=d70,rev=TRUE))
## what matters is (not the mean censoring time but)
## the support of the censoring distribution,
## that is the time region where the probability of
## being uncensored is different from zeroRun the code above in your browser using DataLab