Data Generation : Aprepitant case study (binary endpoint, summary measure as rate difference)

2026-10-03

Source :

set.seed(42)

case_study_config <- yaml::yaml.load_file(system.file("conf/case_studies/aprepitant.yml", package = "BExTE"))
format_case_study_config(case_study_config)

Case Study Configuration
Parameter	Value
General
Name	Aprepitant
Control	Ondansetron
Summary Measure Likelihood	binomial
Theta 0	0
Endpoint	binary
Null Space	left
Sampling Approximation	FALSE
Target
Target Control	52
Target Treatment	57
Target Total	109
Target Responses (Control)	42
Target Responses (Treatment)	48
Target Treatment Effect	0.034412955465587
Target Standard Error	0.072936
Source
Source Control	280
Source Treatment	293
Source Total	573
Source Responses (Control)	154
Source Responses (Treatment)	184
Source Treatment Effect	0.0779863481228669
Source Standard Error	0.04100323

Generation of aggregate data

source_data <- SourceData$new(case_study_config)

drift <- 0.1
target_sample_size_per_arm <- 100

target_data <- TargetDataFactory$new()
target_data <- target_data$create(source_data = source_data, case_study_config = case_study_config, target_sample_size_per_arm = target_sample_size_per_arm, treatment_drift = drift, summary_measure_likelihood = source_data$summary_measure_likelihood)

n_replicates <- 10000
data <- target_data$generate(n_replicates = n_replicates)
print(data[1:10,])

##    sample_control_rate sample_treatment_rate sample_size_per_arm
## 1                 0.47                  0.67                 100
## 2                 0.48                  0.66                 100
## 3                 0.48                  0.75                 100
## 4                 0.51                  0.69                 100
## 5                 0.48                  0.71                 100
## 6                 0.63                  0.73                 100
## 7                 0.56                  0.70                 100
## 8                 0.59                  0.78                 100
## 9                 0.50                  0.71                 100
## 10                0.57                  0.70                 100
##    treatment_effect_estimate treatment_effect_standard_error standard_deviation
## 1                       0.20                      0.06857113          0.6857113
## 2                       0.18                      0.06884766          0.6884766
## 3                       0.27                      0.06611354          0.6611354
## 4                       0.18                      0.06810286          0.6810286
## 5                       0.23                      0.06749074          0.6749074
## 6                       0.10                      0.06558963          0.6558963
## 7                       0.14                      0.06755738          0.6755738
## 8                       0.19                      0.06430397          0.6430397
## 9                       0.21                      0.06752037          0.6752037
## 10                      0.13                      0.06746110          0.6746110

target_data$plot_sample(data)

Implementation details

read_function_code(sample_aggregate_binary_data)

## sample_aggregate_binary_data <- function (rate, n, n_replicates) 
## {
##     n_successes <- rbinom(n_replicates, n, rate)
##     return(n_successes/n)
## }

sample_treatment_rate <- sample_aggregate_binary_data(self$treatment_rate,
                                                              self$sample_size_per_arm,
                                                              n_replicates)

sample_control_rate <- sample_aggregate_binary_data(self$control_rate,
                                                    self$sample_size_per_arm,
                                                    n_replicates)

treatment_effect_standard_error <- sqrt(
  sample_treatment_rate * (1 - sample_treatment_rate) / self$sample_size_control + sample_control_rate * (1 - sample_control_rate) / self$sample_size_treatment
)

standard_deviation <- sqrt(
  sample_treatment_rate * (1 - sample_treatment_rate) + sample_control_rate * (1 - sample_control_rate)
)