In this talk, we will introduce two Bayesian design and analysis methods for phase II clinical trials. First of all, we propose a Bayesian model-assisted adaptive design based on the win ratio (WR) statistic. The proposed design uses the joint asymptotic distribution of the WR statistics across interim and final analyses to derive the stopping boundaries without specifying the underlying outcome distribution. The proposed design allows flexible interim monitoring with early stopping for futility, superiority or toxicity and controls the family-wise error rate (FWER) using the graphical testing procedure. Then, we propose a Bayesian dynamic information borrowing prior for longitudinal data with informative dropout. The proposed prior uses a shared-parameter model to handle the informative dropout and apply a mixture prior framework to incorporate historical control data while accounting for possible prior-data conflict.