Bayesian nonparametric analysis of longitudinal studies in the presence of informative missingness
针对纵向临床试验中不可忽略的非单调缺失数据，提出贝叶斯非参数框架，通过辅助工作先验和识别约束进行推断，支持灵活的敏感性分析。
In longitudinal clinical trials, one often encounters missingness that is thought to be nonignorable. Such missingness introduces identifiability issues, resulting in causal effects being nonparametrically unidentified; it is then prudent to conduct a sensitivity analysis to assess how much of the inference is being driven by untestable assumptions needed to identify the effects of interest. We introduce a Bayesian nonparametric framework for conducting inference in the presence of nonignorable, nonmonotone missingness. This framework focuses on the specification of an auxiliary working prior on the space of complete data generating mechanisms. This prior induces a prior on the observed data generating mechanism, which is then used in conjunction with an identifying restriction to conduct inference. Advantages of this approach include a flexible modelling framework, access to simple computational methods, strong theoretical support, straightforward sensitivity analysis, and applicability to nonmonotone missingness.