Study-Lifecycle Sandbox runs / synthetic

One synthetic early-phase clinical-pharmacology study, taken end-to-end through the analysis tools — synthetic ADaM → PK/NCA → TLF computations — to exercise the whole pipeline on realistically-shaped data. Base R; deterministic; runs offline.

What it is

A self-contained generator (simulate_adam.R) emits eight ADaM-shaped domains — ADSL ADEX ADPC ADPP ADAE ADVS ADLB ADEG — with a one-compartment oral PK model + NCA, a 2×2 bioequivalence crossover option, and per-period safety domains. run_lifecycle.R then runs the demographics, the PK-NCA (geometric mean, log scale), and the bioequivalence ANOVA (90% CI) computations on the simulated study.

100% synthetic

The design structure (design-model mix, enrollment-N distribution, healthy-volunteer rate, randomization/masking, archetype mix, unit geography) is informed by public early-phase Phase-1 registration patterns on ClinicalTrials.gov. All participant-level data is simulated from generic domain priors (typical oral small-molecule PK, healthy-volunteer AE incidence) — it is not fit to any specific sponsor, and no real participant-level data is used or reproduced. This demonstrates the tools and the life cycle, not any real study's outcomes. The public-registry calibration corpus that informed the priors is not included here; the priors it produced are baked into priors.R, so the generator runs standalone.

Run it

Rscript sim_lifecycle/run_lifecycle.R

Source & output

simulate_adam.Rsynthetic ADaM generator (base R, priors baked in) priors.Rdesign + domain priors (hardcoded list) run_lifecycle.Rdriver: simulate → demographics / PK-NCA / BE-ANOVA out/adsl.csvone-record-per-participant (ADSL) out/adpp.csvPK parameters (NCA) out/adae.csvadverse events

Full generated study (seed 2026): the eight out/*.csv ADaM domains. Output feeds the SAS and R TLF pseudocode libraries.