R/gLRT1.R

In glrt: Generalized Logrank Tests for Interval-censored Failure Time Data

gLRT1 <-
function(A, k=2, M=50, EMstep=TRUE, ICMstep=TRUE, tol=1e-06, maxiter=1000, inf=Inf)
{
A[A[,2]==inf,2] = Inf
if(ncol(A) == 3 && all(A[,2] >= A[,1]) && length(unique(A[,3])) == k && all(A[,3]>=0) && all(A[,3]< k) )
{
AA = A[,-3] 
trt = A[,3]
est = ModifiedEMICM(AA, EMstep=EMstep, ICMstep=ICMstep, tol=tol, maxiter=maxiter)
tiny = .Machine$double.eps*100
est$sigma = ifelse(abs(est$sigma) < tiny, 0, est$sigma)
est$sigma = ifelse(abs(1.0 - est$sigma) < tiny, 1.0, est$sigma)
cens = CensorType(A, inf)
u = Teststat1(trt, k, est, cens)
var = Var1(A, est, trt, cens, k=k, M=M)
chisq = u[1:k-1] %*% solve(var[1:k-1, 1:k-1]) %*% u[1:k-1]
p = 1-pchisq(chisq, k-1)
}
else
{
stop("Please Verify data format, # of samples, and treatment indicator!")
}
out = data.frame()
class(out) = "glrt1"
out$method = "Generalized log-rank test (Zhao and Sun, 2004)"
out$u = u
out$var = var
out$chisq = chisq
out$df = k-1
out$p = p
out
}