Adaptive Trial Designs - Alex Kaizer @ ERD Conference 6.5.19

[@juanete69]

Do we need to use cross-validation in bayesian analysis to control overfitting?

[@juanete69]

How do you specify a MAMS model in R with the lme4 library (or any alternative)? (dropping arms and taking into account the correction of type-I errors? How do you compensate the error type-I when different arms have different duration?

[@willnotburn]

Great stuff! Question: why is adaptive design a new thing? Bayes Theorem was formulated in the 18th century. Why has the so-called traditional trial design not been adaptive all along?

[@juanete69]

Because bayesian analysis for real life problems is computationally very expensive, and it has needed the development of MCMC methods.