AdaptiveRandomisation {BayesianPlatformDesignTimeTrend}

AdaptiveRandomisation

Description

This is a function doing the randomisation process. This Function generates the Sequence for patient allocation to each arm, patient outcomes.

Usage

AdaptiveRandomisation(
  Fixratio,
  rand.algo,
  K,
  n.new,
  randomprob,
  treatmentindex,
  groupwise.response.probs,
  group,
  armleft,
  max.deviation,
  trend_add_or_multip,
  trend.function,
  trend.effect,
  ns,
  Fixratiocontrol
)

Arguments

Fixratio

A indicator TRUE/FALSE

rand.algo

Randomisation algorithm: "Coin": Biased coin; "Urn": Urn method

K

Total number of arms at the beginning

n.new

The cohort size

randomprob

A named vector of randomisation probability to each arm

treatmentindex

The vector of treatment arm index excluding the control arm whose index is 0

groupwise.response.probs

A matrix of response probability of each arm

group

The current stage

armleft

The number of treatment left in the platform (>2)

max.deviation

Tuning parameter of using urn randomisation method.

trend_add_or_multip

How time trend affects the true response probability: "add" or "mult"

trend.function

The function returns time trend effect regarding to different time trend pattern

trend.effect

The strength of time trend effect as a parameter in trend.function()

ns

A vector of accumulated number of patient at each stage

Fixratiocontrol

A numeric value indicating the weight of control in randomisation. Eg. 1 means equal randomisation, 2 means thw number of patients allocated to control is twice as large as other treatment arm.

Value

A list of patient allocation and patient outcome nstage: A vector of the number of patients allocated to each arm ystage: A vector of the patients outcome after treating with each arm znew: A vector of treatment index assigned to each patient in the current cohort ynew: A vector of outcome index record for each patient after treatment in the current cohort

Author(s)

Ziyan Wang

References

Mass weighted urn design—a new randomization algorithm for unequal allocations. Zhao, Wenle. Contemporary clinical trials 43 (2015): 209-216.

Examples

AdaptiveRandomisation(
Fixratio = FALSE,
rand.algo = "Urn",
K = 2,
n.new = 30,
randomprob = matrix(c(0.5, 0.5), ncol = 2, dimnames = list(c(),c("1","2"))),
treatmentindex = 1,
groupwise.response.probs = matrix(rep(c(0.4, 0.4), 5), byrow = TRUE, ncol = 2, nrow = 5),
group = 1,
armleft = 2,
max.deviation = 3,
trend_add_or_multip = "mult",
trend.function = function(ns, group, i, trend.effect) {delta = 0; return(delta)},
trend.effect = c(0, 0),
ns = c(30, 60, 90, 120, 150),
Fixratiocontrol = NA)

AdaptiveRandomisation(
Fixratio = TRUE,
rand.algo = "Urn",
K = 4,
n.new = 30,
randomprob = NA,
treatmentindex = c(1,3),
groupwise.response.probs = matrix(rep(c(0.4, 0.4,0.4, 0.4), 5), byrow = TRUE, ncol = 4, nrow = 5),
group = 1,
armleft = 3,
max.deviation = 3,
trend_add_or_multip = "mult",
trend.function = function(ns, group, i, trend.effect) {delta = 0; return(delta)},
trend.effect = c(0, 0),
ns = c(30, 60, 90, 120, 150),
Fixratiocontrol = 1)