minimizePocSim: use the Pocock-Simon or Taves algorithm for computing...

In randPack: Randomization routines for Clinical Trials

Description

use the Pocock-Simon or Taves algorithm for computing covariate-adaptive 'minimization' allocations for a clinical trial

Usage

minimizePocSim(df, features, trtvec, obsdf, trttab, f = function(x, y) sum(abs(x + 1 - y)))
minimizeTaves(df, features, trtvec, obsdf, trttab)

Arguments

df	a data frame with columns corresponding to covariates rows corresponding to subjects
features	character vector of covariates to use
trtvec	vector of assignments made so far
obsdf	data frame for incoming observation, with values for all components enumerated in features
trttab	table of treatment ratios
f	score that determines impending allocation

Details

These functions are generally not called directly. See the vignette; if supplied as the method slot of a MinimizationDesc object the appropriate data are assembled as arguments to these functions.

Value

a treatment code

Examples

new("MinimizationDesc", treatments=c(A=1L, B=1L), method=minimizePocSim,
   type="Minimization", featuresInUse="sex")

Example output

An object of class "MinimizationDesc"
Slot "method":
function (df, features, trtvec, obsdf, trttab, f = function(x, 
    y) sum(abs(x + 1 - y))) 
{
    picks = factorCounts(df, features, trtvec, obsdf)
    if (length(picks) > 2) 
        stop("only works for two-arm studies")
    if (length(picks) < 2) {
        return(sample(rep(names(trttab), times = trttab), size = 1))
    }
    sco1 = f(picks[[1]], picks[[2]])
    if (length(sco1) != 1 || !is.numeric(sco1)) 
        stop("f must return numeric scalar")
    sco2 = f(picks[[2]], picks[[1]])
    ans = names(picks)
    ifelse(sco1 > sco2, ans[2], ans[1])
}
<environment: namespace:randPack>

Slot "featuresInUse":
[1] "sex"

Slot "treatments":
A B 
1 1 

Slot "type":
[1] "Minimization"

randPack documentation built on Nov. 8, 2020, 5:32 p.m.