simulate_rr: Restricted randomization

In yevgenryeznik/artool: Adaptive Randomization Routine for Clinical Trials

Description

Makes multiple simulates of a clinical trial with retricted randomizaion procedures

Usage

simulate_rr(nsim, nsbj, w, proc, proc_param, distr, distr_param)

Arguments

nsim	number of simulation runs.
nsbj	number of subjects to randomize.
w	vector of fixed allocation ratio (integers with GCD = 1).
proc	name of randomization procedure c("CRD", "PBD", "BUD", "MWUD", "DBCD", "DL", "MinQD", "MaxEnt").
proc_param	tuning parameter of randomization procedure (NA for "CRD").
distr	distribution of response c("binary", "uniform", "normal", "exponential", "weibull", "loglogistic", "lognormal"). See response for details.
distr_param	parameter(s) of the response distribution (a list with numeric vector(s) with the same length as w). See response.

Value

a list with three items:

• treatment – a matrix of treatment assignments with nsim rows and nsbj columns, i.e. r-th row contains tretament assignments of r-th simulation,

• response – a matrix of responses with nsim rows and nsbj columns, i.e. r-th row contains responses of r-th simulation,

• op – a data frame which contains operating characteristics:

  • MI – maximum imbalance,

  • AFI – average forcing index,

  • AMPM1 – average momentum of probability mass based on expected predicted imbalance,

  • ACMPM1 – average cumulative momentum of probability mass based on expected predicted imbalance,

  • AMPM2 – average momentum of probability mass based on the expected imbalance,

  • ACMPM2 – average cumulative momentum of probability mass based on expected predicted imbalance,

  • ASD – average standard deviation of allocation proportions,

• probability – a data frame with unconditional allocation probabilities vs number of subjects for each treatment;

• allocation – a data frame with allocation proportions in the end of a trial vs number of simulations for each treatment (could be usefull to make a boxplot of allocation proportion distribution).

Author(s)

Yevgen Ryeznik ([e-mail]), Oleksandr Sverdlov

Examples

   simulate_rr(1000, 200, c(4, 3, 2, 1), "DL", 2, "normal", list(mean = rep(0, 4), sd = rep(1, 4)))
   simulate_rr(1000, 200, c(31, 13), "MaxEnt", 0.5, "normal", list(mean = c(0, 0.2), sd = c(1, 1)))    

yevgenryeznik/artool documentation built on May 15, 2019, 5:35 p.m.