Adaptive Interim Analysis and Multiplicity Control Toolkit

Reproducible R toolkit for adaptive interim monitoring and familywise error control with:

• Closed testing and combination tests (fisher, inverse_normal)
• Group-sequential alpha spending (obrien_fleming, pocock)
• Binary endpoint utilities (two-arm proportion test)
• Time-to-event utilities (stratified log-rank and stratified Cox PH)
• Scenario-driven simulation workflows and TFL-style R Markdown reporting

Project Layout

• R/: core framework APIs and simulation engine
• config/: defaults and fast/deep simulation scenario grids
• analyses/: simulation and report entrypoint scripts
• reports/: parameterized master R Markdown report
• tests/testthat/: unit and QC tests
• outputs/: generated simulation and report artifacts

Quick Start

make bootstrap
make test
make simulate-fast
make render-report

Deep simulation mode:

make simulate-deep

Main APIs

• design_interim(info_frac, alpha, spending, sided, looks)
• boundaries_gs(design)
• combine_pvalues(p_vals, method, weights)
• closed_test(p_matrix, combination_method, hypothesis_set)
• decision_rules(boundaries, combined_results, look_id)
• stat_binary_twoarm(data, effect_scale)
• stat_tte_strat_logrank(data, strata_vars)
• fit_tte_strat_cox(data, strata_vars)
• load_defaults(path)
• load_scenarios(mode, path)
• run_simulation_grid(...)

Reproducibility Defaults

Defaults are centralized in config/defaults.yml:

• one-sided alpha = 0.025
• default looks at information fractions 0.50, 0.75, 1.00
• default combination method inverse_normal
• fixed random seed for deterministic runs

Outputs

• Simulation summaries: outputs/simulations/simulation_summary_<mode>.csv
• Replicate-level output (optional): outputs/simulations/simulation_replicates_<mode>.csv
• Boundary tables: outputs/simulations/boundaries_<mode>.csv
• Rendered report: outputs/reports/interim_report_<mode>.html