simcrm: Simulate CRM design

In raredd/desmon2: More tools for the design and monitoring of clinical trials

Simulate CRM design

Description

Simulate a CRM dose-escalation study.

Usage

simcrm(probs, expansion = 0L, target = 1/6, nptmax = length(probs) * 6L, ...)

Arguments

probs	vector of true rates of DLTs
expansion	(optional) integer giving size of expansion cohort
target	target AE rate
nptmax	max number of patients used on study (not including the expansion cohort)
...	additional arguments passed to simCrm

Value

A list with following components:

$data	UBCRM::simCrm()$data
$level	mtd level (i.e., row of data selected)
$mtd	DLT rate of dose level selected
$expansion	observed rate of DLT in expansion cohort
$cohort	observed rate of DLT in all patients
$enrolled	total number of patients enrolled on study

See Also

sim3p3

Examples

pr <- 1:5 / 10
simcrm(pr)


## run 100 crm studies
sim1 <- replicate(100L, simcrm(pr), simplify = FALSE)

op <- par(mfrow = c(1, 2), las = 1L)
out <- sapply(sim1, function(x) c(x$level, x$enrolled))
barplot(
  prop.table(table(out[1, ])),
  main = 'Dose selected',
  xlab = 'RP2D',
  ylab = 'Proportion selected'
)
boxplot(
  split(out[2, ], out[1, ]),
  ylim = c(0, length(pr) * 6 + 10),
  main = 'N used to select dose x',
  xlab = 'Dose selected\n(0 - no safe dose)',
  ylab = 'Total size'
)
par(op)

raredd/desmon2 documentation built on April 13, 2025, 1:56 a.m.