Topic: Thompson Sampling for Trials

Deriving the Normal-Normal Posterior for Allocation

Deriving the Thompson Sampling Allocation Identity

Defining the Best Treatment Through Clinical Utility

Conditional Assignment Probabilities Given Trial History

Checking Posterior Best Probabilities by Numerical Integration

Choosing Priors for Treatment Response Rates

Batch Assignment From a Shared Posterior State

Best-Arm Allocation Versus a Clinical Threshold

Best-Arm Probability Versus Predictive Response Probability

Beta-Bernoulli Updating for Binary Outcomes