ABSTRACT
To address the concern with a more limited borrowing to reduce the chance of overinfluence from the prior, Bayesian dynamic borrowing is a statistical approach to account for the inconsistency between the prior datasets and current study by learning how much information to be borrowed. Since Ibrahim and Chen proposed their landmark idea of power prior, various approaches have been proposed for dynamic borrowing in the literature over the last two decades, such as the modified power prior, the commensurate prior, and the adaptive power prior using empirical Bayesian. Several recent FDA approvals of medical devices have included the use of Bayesian dynamic borrowing as briefly summarized in the EEM Framework by MDIC. One of the advantages of Bayesian dynamic borrowing is to examine the compatibility between the reference and target data so that more data borrowing will occur if the compatibility is high and less or no borrowing will occur if the compatibility is low.
