ewoc_simulation {ewoc}

EWOC simulation

Description

Generic function for simulating EWOC trials.

Usage

ewoc_simulation(
  step_zero,
  n_sim,
  sample_size,
  response_sim,
  fixed_first_cohort = TRUE,
  n_cohort = 1,
  alpha_strategy = "conditional",
  alpha_rate = 0.05,
  stop_rule_sim = NULL,
  ncores = 1,
  seed = 1234,
  ...
)

Arguments

step_zero

an object from the classes either 'ewoc_d1classical' or 'ewoc_d1extended' or 'ewoc_d1ph' created using the first cohort data.

n_sim

a number indicating the number of phase I clinical trials to be simulated.

sample_size

a number indicating the number of patients enrolled for each clinical trial.

response_sim

a function which is self-contained and will be used as a generator function of the response variables in the simulation. Its only input is 'dose' and output is the indicator of DLT for classical and extended EWOC and the time until DLT for proportional hazards EWOC.

fixed_first_cohort

a logical value indicating if the first cohort should be randomly generated or be fixed as the input in 'step_zero'.

n_cohort

a number indicating the number of patients enrolled at each cohort. It is only used for 'ewoc_d1classical' and 'ewoc_d1extended'.

alpha_strategy

a character indicating the strategy to apply for the feasibility value. Default is "constant". Options are "increasing" and "conditional".

alpha_rate

a numerical value indicating the rate of the feasibility strategy. Only necessary if alpha_strategy is either 'increasing' or 'conditional'.

stop_rule_sim

a function having as an input an object containing all the information related to the trial as the returned object trial from either ewoc_d1classical, ewoc_d1extended, ewoc_d1ph and as output a logical value indicating the trial should be stopped.

ncores

a numeric value indicating the number of cores to be used in the simulation performed in parallel. Use parallel::detectCores() to check the number of cores available.

seed

is an integer value, containing the random number generator (RNG) state for random number generation.

...

For an object step_zero with class 'ewoc_d1ph', the argument rate_sim which controls the rate of accrue of patients following a Poisson process. The default is 1.

Value

alpha_sim a matrix n_sim x sample_size containing the values of feasibility used for each step in the trial and each trial in the simulation.

dlt_sim a matrix n_sim x sample_size containing ones and zeros indicating the occurrence of DLT (1) and the absence of DLT (0) for each step in the trial and each trial in the simulation.

dose_sim a matrix n_sim x sample_size containing the doses assigned for each step in the trial and each trial in the simulation.

mtd_sim a numeric vector n_sim x 1 containing the recommended MTD for each trial in the simulation.

rho_sim a numeric vector n_sim x k containing the estimated rho parameter(s) for each trial in the simulation, where k = 1 for ewoc_d1classical, ewoc_d1ph, and k = 2 for ewoc_d1extended.

Examples

## Not run: 
### Classical EWOC
DLT <- 0
dose <- 20
step_zero <- ewoc_d1classical(DLT ~ dose, type = 'discrete',
                          theta = 0.33, alpha = 0.25,
                          min_dose = 20, max_dose = 100,
                          dose_set = seq(20, 100, 20),
                          rho_prior = matrix(1, ncol = 2, nrow = 1),
                          mtd_prior = matrix(1, ncol = 2, nrow = 1),
                          rounding = "nearest")
response_sim <- response_d1classical(rho = 0.05, mtd = 60, theta = 0.33,
                                 min_dose = 20, max_dose = 100)
sim <- ewoc_simulation(step_zero = step_zero,
                      n_sim = 2, sample_size = 30, n_cohort = 1,
                      alpha_strategy = "conditional",
                      response_sim = response_sim,
                      ncores = 1)

### Extended EWOC
DLT <- 0
dose <- 20
step_zero <- ewoc_d1extended(DLT ~ dose, type = 'discrete',
                          theta = 0.33, alpha = 0.25,
                          min_dose = 20, max_dose = 100,
                          dose_set = seq(20, 100, 20),
                          rho_prior = matrix(1, ncol = 2, nrow = 2),
                          rounding = "nearest")
response_sim <- response_d1extended(rho = c(0.05, 0.5),
                                  min_dose = 20, max_dose = 100)
sim <- ewoc_simulation(step_zero = step_zero,
                      n_sim = 2, sample_size = 30, n_cohort = 1,
                      alpha_strategy = "conditional",
                      response_sim = response_sim,
                      ncores = 1)

### PH EWOC
time <- 0
status <- 0
dose <- 20

step_zero <- ewoc_d1ph(cbind(time, status) ~ dose, type = 'discrete',
                     theta = 0.33, alpha = 0.25, tau = 10,
                     min_dose = 20, max_dose = 100,
                     dose_set = seq(20, 100, 20),
                     rho_prior = matrix(1, ncol = 2, nrow = 1),
                     mtd_prior = matrix(1, ncol = 2, nrow = 1),
                     distribution = 'exponential',
                     rounding = 'nearest')
response_sim <- response_d1ph(rho = 0.05, mtd = 60, theta = 0.33,
                            min_dose = 20, max_dose = 100,
                            tau = 10, distribution = "exponential")
sim <- ewoc_simulation(step_zero = step_zero,
                      n_sim = 2, sample_size = 30, n_cohort = 1,
                      alpha_strategy = "conditional",
                      response_sim = response_sim,
                      ncores = 1)

## End(Not run)