ELYP (version 0.7-1)

simuDataYP: Generate random times that follow the YP model with the Given Parameters th1, th2, and alphaX.

Description

This function is for simulations. It generate data with given/known parameters and may be used later to see how well some estimation procedure works on them. th1 = exp(beta1), th2 = exp(beta2), alphaX = a^T X. There is always a covariate Z that indicate the two samples, and the hazards of the two treatments follows YP model. The baseline hazard of sample one (where Z=0) is taken to be exponential.

Usage

simuDataYP(n1, n2, th1, th2, cens, alphaX)

Arguments

n1

sample size of first arm.

n2

sample size of second arm.

th1

the parameter of th1=exp(beta1). Short term.

th2

the parameter of th2=exp(beta2). Long term.

cens

logical, Either TRUE or FALSE.

alphaX

a vector of length n1+n2. It is the inner product of alpha and covariates X....the part that is proportional hazard. This way, alpha can be p dimensional, However alpha times X is always a vector of length n1+n2.

Value

• A list with the following components:
• Y
  The survival times, possibly right censored.
• d
  The censoring status.
• Zmat
  the covariates used in generating random times.

Details

The hazard of the generated survival times, Y, have hazard function that is proportional to exp( alphaX ). The hazard of arm 1 is constant, just exp( alphaX ). The hazard of arm 2 is given as exp(alphaX) / [ 1/th1 S_0(t) + 1/th2 F_0(t) ] where S_0 and F_0 are survival function and CDF of a standard exponential random variable.

References

Yang and Prientice. (2005). Biometrika

Examples

## generate data and covariates.
X <- -99:100/50        ## the covariate for alpha, 200 long
temp <- simuDataYP(n1=100, n2=100, th1=exp(1), th2=exp(-1), cens=TRUE, alphaX = -0.5*X)
## this generate a sample of censored data with n=200. beta1=1, beta2=-1, alpha= -0.5.
## and the design matrix or covariance matrix is 
Zmat <- cbind(X, temp$Zmat)