Bayesian Analysis Simulator

Tools for Bayesian clinical trial design informed by current regulatory methodology guidance for IND/NDA/BLA submissions. Compute posterior distributions, assess prior sensitivity, and evaluate operating characteristics under conjugate models.

1 Model Specification

Endpoint Type

Prior Type

Noninformative

Jeffreys prior: Beta(0.5, 0.5) for binary; Normal(0, 100) for continuous. Lets the data dominate.

Skeptical

Centered at the null: Beta(1, 1) for binary; Normal(0, 1) for continuous. Requires strong evidence to shift.

Informative

User-specified parameters from historical data or expert elicitation.

α (Beta prior)

β (Beta prior)

2 Observed Data

Sample Size (N)

Observed Successes (x)

3 Success Criterion

Threshold a (min meaningful effect) ?

Probability threshold c ?

Prior vs Posterior Distribution

Cumulative Posterior

Prior Contribution

1 Observed Data & Criterion

Endpoint Type

Sample Size (N)

Observed Successes (x)

Threshold a

Probability threshold c

2 Informative Prior Parameters

α (Informative Beta)

β (Informative Beta)

3 Prior Comparison

Prior Type	Prior Params	Posterior Mean	95% CrI	Pr(θ>a)	Decision

Posterior Distributions Under Different Priors

1 Design Parameters

Endpoint Type

True Response Rate (design prior) ?

Null Hypothesis Value

Min Sample Size

Max Sample Size

Step Size

Analysis Prior

Success Threshold a

Probability Threshold c

Simulations per N ?

Power Curve (Under Alternative)

Type I Error (Under Null)

2 Summary Table

N	Power	Type I Error	E[Posterior Width]