Bayesian Toolkit
The Bayesian Toolkit is a suite of calculators for Bayesian clinical trial design and monitoring. It is not a single calculation — each module addresses a distinct task, from encoding a prior to sizing single- and two-arm trials, borrowing historical data, projecting interim success, and designing sequential monitoring.
How it works
The modules are designed to be used together: elicit a prior, optionally borrow historical control data, then size a single-arm or two-arm design, and plan interim decisions with predictive probability or posterior-probability sequential monitoring. The modules use compatible Bayesian models and outputs. The trial-design and monitoring modules estimate operating characteristics by simulation, while prior elicitation and borrowing modules characterize the prior information supplied downstream. Each module has its own dedicated calculator and documentation.
When to use it
- You are planning a Bayesian trial and need prior elicitation, sample size, borrowing, or interim monitoring tools that share a consistent approach.
- You want posterior- or predictive-probability decision rules with simulated operating characteristics rather than p-values alone.
- You are exploring which Bayesian design module fits your endpoint and objective.
Assumptions & limitations
- The toolkit is an overview and launch point; each design decision (prior, sample size, borrowing, monitoring) is made in its own module with its own assumptions.
- Across the modules, decision thresholds are posterior or predictive probabilities — the frequentist type I error and power are separately estimated by simulation, not implied by the threshold.
- The conjugate models and simulated operating characteristics describe expected behavior under the modeled assumptions, not guarantees for a realized trial.
For the full methodology, derivation, and worked examples, see the guide for your design: