A Bayesian method to measure predictive performance of a model in a different population from that used to develop the model in clinical development for pediatric or rare diseases