assessDesign — assessDesign • BayesianMCPMod

This function performs simulation based trial design evaluations for a set of specified dose-response models

Usage

assessDesign(
  n_patients,
  mods,
  prior_list,
  sd = NULL,
  contr = NULL,
  dr_means = NULL,
  data_sim = NULL,
  estimates_sim = NULL,
  n_sim = 1000,
  alpha_crit_val = 0.05,
  modeling = FALSE,
  simple = TRUE,
  avg_fit = TRUE,
  reestimate = FALSE,
  delta = NULL,
  evidence_level = NULL,
  n_bs_samples = 1000,
  med_selection = c("avgFit", "bestFit"),
  probability_scale = FALSE
)

Arguments

n_patients: Vector specifying the planned number of patients per dose group. A minimum of 2 patients are required in each group.
mods: An object of class Mods as specified in the DoseFinding package.
prior_list: A prior_list object specifying the utilized prior for the different dose groups
sd: A positive value, specification of assumed sd. Not required if either data_sim or estimates_sim is provided. Also not required in case of binary endpoint. Default NULL
contr: An object of class optContr as created by the DoseFinding::getContr function. Allows specification of a fixed contrasts matrix. Default NULL.
dr_means: A vector, allows specification of individual (not model based) assumed effects per dose group. Default NULL.
data_sim: An optional data frame for custom simulated data. Must follow the data structure as provided by simulateData(). Default NULL.
estimates_sim: An optional named list of 1) list of vectors for the estimated means per dose group (estimates_sim$mu_hats) and 2) of list of matrices for the covariance matrices specifying the (estimated) variabilities (estimates_sim$S_hats). Dimensions of entries must match the number of dose levels. Default NULL.
n_sim: Number of simulations to be performed
alpha_crit_val: (Un-adjusted) Critical value to be used for the MCP testing step. Passed to the getCritProb function for the calculation of adjusted critical values (on the probability scale). Default 0.05.
modeling: Boolean variable defining whether the Mod part of Bayesian MCP-Mod will be performed in the assessment. More heavy on resources. Default FALSE.
simple: Boolean variable defining whether simplified fit will be applied, see ?getModelFits. Set automatically to TRUE if argument delta is provided. Passed to getModelFits Default TRUE.
avg_fit: Boolean variable, defining whether an average fit (based on generalized AIC weights) should be performed in addition to the individual models. Default TRUE.
reestimate: Boolean variable defining whether critical value should be calculated with re-estimated contrasts (see getCritProb function for more details). Default FALSE.
delta: A numeric value for the threshold Delta for the MED assessment. If NULL, no MED assessment is performed. Default NULL.
evidence_level: A numeric value between 0 and 1 for the evidence level gamma for the MED assessment. Only required for Bayesian MED assessment, see ?getMED for details. If NULL, MED assessment will be performed on the fitted model according to the argument med_selection. Default NULL.
n_bs_samples: Number of bootstrap samples for the MED assessment if evidence_level is provided. Note that more extreme quantiles (i.e., quantiles closer to 0 or 1) tend to require more bootstrap samples to maintain precision. Default 1000.
med_selection: A string, either "avgFit" or "bestFit", for the method of MED selection. Default "avgFit".
probability_scale: A boolean to specify if the trial has a continuous or a binary outcome. Setting to TRUE will transform calculations from the logit scale to the probability scale, which can be desirable for a binary outcome. Default FALSE.

Value

Returns success probabilities for the different assumed dose-response shapes, attributes also includes information around average success rate (across all assumed models) and prior Effective sample size.

Examples

mods <- DoseFinding::Mods(linear      = NULL,
                          emax        = c(0.5, 1.2),
                          exponential = 2,
                          betaMod     = c(1, 1),
                          doses       = c(0, 0.5, 2,4, 8),
                          maxEff      = 6)
                          
sd <- 12
prior_list <- list(Ctrl = RBesT::mixnorm(comp1 = c(w = 1, m = 0, s = 12), sigma = 2),
                   DG_1 = RBesT::mixnorm(comp1 = c(w = 1, m = 1, s = 12), sigma = 2),
                   DG_2 = RBesT::mixnorm(comp1 = c(w = 1, m = 1.2, s = 11), sigma = 2) ,
                   DG_3 = RBesT::mixnorm(comp1 = c(w = 1, m = 1.3, s = 11), sigma = 2) ,
                   DG_4 = RBesT::mixnorm(comp1 = c(w = 1, m = 2, s = 13), sigma = 2))
n_patients <- c(40, 60, 60, 60, 60)
dose_levels  <- c(0, 0.5, 2, 4, 8)

success_probabilities <- assessDesign(
  n_patients  = n_patients,
  mods        = mods,
  prior_list  = prior_list,
  sd          = sd,
  n_sim       = 1e2) # speed up example run time

success_probabilities
#> $linear
#> Bayesian Multiple Comparison Procedure
#>   Estimated Success Rate: 0.86 
#>   N Simulations:          100 
#>    Model Shape:         lin emax1 emax2   exp betaM 
#>    Significance Freq:  0.81  0.60  0.75  0.74  0.31 
#> 
#> $emax1
#> Bayesian Multiple Comparison Procedure
#>   Estimated Success Rate: 0.88 
#>   N Simulations:          100 
#>    Model Shape:         lin emax1 emax2   exp betaM 
#>    Significance Freq:  0.57  0.81  0.78  0.24  0.74 
#> 
#> $emax2
#> Bayesian Multiple Comparison Procedure
#>   Estimated Success Rate: 0.9 
#>   N Simulations:          100 
#>    Model Shape:         lin emax1 emax2   exp betaM 
#>    Significance Freq:  0.70  0.81  0.83  0.43  0.79 
#> 
#> $exponential
#> Bayesian Multiple Comparison Procedure
#>   Estimated Success Rate: 0.85 
#>   N Simulations:          100 
#>    Model Shape:         lin emax1 emax2   exp betaM 
#>    Significance Freq:  0.81  0.33  0.54  0.83  0.04 
#> 
#> $betaMod
#> Bayesian Multiple Comparison Procedure
#>   Estimated Success Rate: 0.86 
#>   N Simulations:          100 
#>    Model Shape:         lin emax1 emax2   exp betaM 
#>    Significance Freq:  0.28  0.61  0.67  0.04  0.85 
#> 
#> attr(,"avgSuccessRate")
#> [1] 0.87
#> attr(,"placEff")
#> [1] 0
#> attr(,"maxEff")
#> [1] 6
#> attr(,"sampleSize")
#> [1] 40 60 60 60 60
#> attr(,"priorESS")
#> Ctrl DG_1 DG_2 DG_3 DG_4 
#>    0    0    0    0    0 

if (interactive()) { # showcasing further functionality

## Analysis with custom data
data_sim <- simulateData(
  n_patients        = n_patients,
  dose_levels       = dose_levels,
  sd                = sd,
  mods              = mods,
  n_sim             = 10)

success_probabilities_cd <- assessDesign(
  n_patients  = n_patients,
  mods        = mods,
  prior_list  = prior_list,
  data_sim    = data_sim,
  sd          = sd,
  n_sim       = 1e2) # speed up example run time

success_probabilities_cd

## Analysis with custom dose response relationship
custom_dr_means <- c(1, 2, 3, 4, 5)

success_probs_custom_dr <- assessDesign(
  n_patients  = n_patients,
  mods        = mods,
  prior_list  = prior_list,
  dr_means    = custom_dr_means,
  sd          = sd,
  n_sim       = 1e2) # speed up example run time

success_probs_custom_dr

## Analysis with custom estimates for means and variabilies
## No simulated data, only simulated model estimates
estimates_sim <- list(mu_hats = replicate(100, list(c(1, 2, 3, 4, 5) + rnorm(5, 0, 1))),
                      S_hats  = list(diag(1, 5)))

success_probs_custom_est <- assessDesign(
  n_patients    = n_patients,
  mods          = mods,
  prior_list    = prior_list,
  estimates_sim = estimates_sim)

success_probs_custom_est

}

if (interactive()) { # takes typically > 5 seconds

# with MED estimation without bootstrapping
# see ?getMED for details

success_probabilities <- assessDesign(
  n_patients     = n_patients,
  mods           = mods,
  prior_list     = prior_list,
  sd             = sd,
  modeling       = TRUE,
  n_sim          = 10, # speed up example run time
  delta          = 7)

  success_probabilities

# with MED estimation with bootstrapping

success_probabilities <- assessDesign(
  n_patients     = n_patients,
  mods           = mods,
  prior_list     = prior_list,
  sd             = sd,
  modeling       = TRUE,
  n_sim          = 10, # speed up example run time
  delta          = 7,
  evidence_level = 0.8)

  success_probabilities

}