Clinical trials using Bayesian adaptive design have been proposed by medical device or drug companies. In this type of study design, Bayesian statistics are used to determine early stop for success/futility or continuation to next interim/final look. The type-1 error in such a design has been difficult to be assessed. One of the advantages of the design is that the to-be-observed values can be imputed by sampling from posterior distribution in the calucatlion of Bayesian predictive probability of success (PPS). We will discuss the pros and cons of Bayesian adaptive design and the expected key elements needed for adequate regulatory review.