Phase II–III design lab.
Find a signal. Test a claim. Watch the evidence change.
Change the fictional world Reveal the assumptions.
Rule, calculation & limits Trace the decision.
Fictional simulation data View or export this look.
Complete design map & sources 56 runnable examples.
A trial can combine several of these choices. “Parallel” describes allocation; “Bayesian” describes inference; “decentralized” describes conduct. Phase tags indicate common uses, not regulatory definitions.
Specified rules implement the stated mathematical model. Mechanism demos simplify named approaches. Study formats illustrate structure or conduct. These examples do not rank published methods or replace clinical, statistical and ethical protocol review.
Current Bayesian & adaptive context Checked October 2026.
The FDA’s January 2026 Bayesian guidance and June 2026 master-protocol revision are drafts, not final guidance.
Recent examples include BOP2-TE (online 2024; issue 2026) and similarity-weighted external borrowing (2026). Their cards demonstrate mechanisms; they do not reproduce the published calibrated algorithms.