Generate simulation data to test time-dependent Cox model with adaptive lasso

Source: R/simulation.R

SIM.FUN.TVC generates time-dependent survival response with four time-intervals 0-1, 1-2, 2-3, 3-4 for each subject data. All subjects are administratively censored at 4, if T>4. T comes from a Weibull distribution with shape of 2. The design matrix comes from a correlated multivariate normal. The default signals (beta0) are sparse.

SIM.FUN.TVC(p.ti = 50, p.tv = 50, n.subject = 1e+06, cor = 0.2,
  beta0.ti = NULL, beta0.tv = NULL)

Arguments

p.ti

number of time-invariant covariates
p.tv

number of time-varying covariates
n.subject

number of subjects
cor

correlation between time-varying and each interval's time-varying covairates
beta0.ti

coefficient for time-invariant covariates
beta0.tv

coefficients for time-varying covariates

Value

a matrix with the first column starting time, second column ending time, third column event (0,1), and rest = design matrix + ID for subject.

References

Section 3.3 in Austin, P.C., 2012. Generating survival times to simulate Cox proportional hazards models with time‐varying covariates. Statistics in medicine, 31(29), pp.3946-3958.

Author

Yan Wang, Tianxi Cai, Chuan Hong