sim_rct_biomarker: Simulate a randomized clinical trial with biomarkers

In smog: Structural Modeling by using Overlapped Group Penalty

Description

sim_rct_biomarker is used to simulate clinical trial data with specified treatment, prognostic, and predictive effect sizes.

Usage

sim_rct_biomarker(n = 50, p = 100, p_prog = 5, p_pred = 5,
  p_both = 5, v_trt = 0.4, v_prog = 0.2, v_pred = 0.2,
  v_err = 0.2, corr = NULL, family = "gaussian", ...)

Arguments

n	Number of subjects.
p	Number of biomarkers.
p_prog	Number of biomarkers with prognostic effects only.
p_pred	Number of biomarkers with predictive effects only.
p_both	Number of biomarkers with both prognostic and predictive effects
v_trt	Variance of response due to treatment.
v_prog	Variance of response due to prognostic effects.
v_pred	Variance of response due to predictive effects.
v_err	Variance of response due to random noise.
corr	Autocorrelation parameter between biomarkers, default is NULL.
family	The distribution family for response variable, can be gaussian'', or binomial”. Default is “gaussian”.
...	further arguments passed to or from other methods.

Value

A list containing several variables.

T

Treatment status in 1 or -1 values.

X

Biomarkers.

W

Hadamard product of treatment and biomarkers.

M

Model matrix - binding of T, X, and W.

Y

Response.

Y0

Response without error.

tau

Treatment effect.

beta

Prognostic effects.

gamma

Predictive effects.

theta

All effects corresponding to M.

Author(s)

Chong Ma [e-mail], Kevin Galinsky [e-mail].

References

\insertRef

ma2019structuralsmog

Examples

sim <- sim_rct_biomarker(n = 1e3)
var(as.vector(sim$T * sim$tau))
var(as.vector(sim$X %*% sim$beta))
var(as.vector(sim$W %*% sim$gamma))

smog documentation built on Aug. 10, 2020, 5:07 p.m.