generate_bootstrap_synthetic

Generate Synthetic Data using Bootstrap with Perturbation

internal

Implements statistical methods for exploratory subgroup identification
in clinical trials with survival endpoints. Provides tools for identifying
patient subgroups with differential treatment effects using machine learning
approaches including Generalized Random Forests (GRF), LASSO regularization,
and exhaustive combinatorial search algorithms. Features bootstrap bias
correction using infinitesimal jackknife methods to address selection bias
in post-hoc analyses. Designed for clinical researchers conducting exploratory
subgroup analyses in randomized controlled trials, particularly for
multi-regional clinical trials (MRCT) requiring regional consistency
evaluation. Supports both accelerated failure time (AFT) and Cox proportional
hazards models with comprehensive diagnostic and visualization tools.
Methods are described in León et al. (2024) <doi:10.1002/sim.10163>.

Larry Leon

forestsearch

Exploratory Subgroup Identification in Clinical Trials with
Survival Endpoints

generate_bootstrap_synthetic function

<dl><dt>data</dt>
<dd>Original dataset to bootstrap from</dd>
<dt>continuous_vars</dt>
<dd>Character vector of continuous variable names</dd>
<dt>cat_vars</dt>
<dd>Character vector of categorical variable names</dd>
<dt>n</dt>
<dd>Number of synthetic observations to generate (default: same as original)</dd>
<dt>seed</dt>
<dd>Random seed for reproducibility</dd>
<dt>noise_level</dt>
<dd>Noise level for perturbation (0 to 1, default 0.1)</dd>
<dt>id_var</dt>
<dd>Optional name of ID variable to regenerate (will be numbered 1:n)</dd>
<dt>cat_flip_prob</dt>
<dd>Probability of flipping categorical values (default: noise_level/2)</dd>
<dt>preserve_bounds</dt>
<dd>Logical: should continuous variables stay within original bounds? (default: TRUE)</dd>
<dt>ordinal_vars</dt>
<dd>Optional character vector of ordinal categorical variables
(these will be perturbed to adjacent values rather than randomly flipped)</dd></dl>

Arguments

Generate Synthetic Data using Bootstrap with Perturbation — generate_bootstrap_synthetic

<dl>

<dt>data</dt>
<dd>Original dataset to bootstrap from</dd>


<dt>continuous_vars</dt>
<dd>Character vector of continuous variable names</dd>


<dt>cat_vars</dt>
<dd>Character vector of categorical variable names</dd>


<dt>n</dt>
<dd>Number of synthetic observations to generate (default: same as original)</dd>


<dt>seed</dt>
<dd>Random seed for reproducibility</dd>


<dt>noise_level</dt>
<dd>Noise level for perturbation (0 to 1, default 0.1)</dd>


<dt>id_var</dt>
<dd>Optional name of ID variable to regenerate (will be numbered 1:n)</dd>


<dt>cat_flip_prob</dt>
<dd>Probability of flipping categorical values (default: noise_level/2)</dd>


<dt>preserve_bounds</dt>
<dd>Logical: should continuous variables stay within original bounds? (default: TRUE)</dd>


<dt>ordinal_vars</dt>
<dd>Optional character vector of ordinal categorical variables
(these will be perturbed to adjacent values rather than randomly flipped)</dd>

</dl>

generate_bootstrap_synthetic: Generate Synthetic Data using Bootstrap with Perturbation

Description

Usage

Value

Arguments