Statistical Inferences For Incomplete Data With Missing And Truncation In Clinical Trials

We discuss how to deal with the classified incomplete data with missing and truncation in clinical trials mainly in the paper. In the clinical research process, as a result of some kind of reason, there might inevitably appear the situation with incomplete data. This article mainly takes Tumor Xenograft Model as example, which experimental mice are grafted with human cancer cells, to discuss the inference for the classified incomplete data with missing and truncation in clincal trials. In cancer drug development, demonstrating activity in xenograft models is an important step in bringing a promising compound to humans, in which, a key outcome variable is the tumor volume measured in a given period of time for groups of mice given diffent doses of cancer drug. However, for some reasons, a mouse may die before the end of the trial, or may lose the nessary to go on the trial when its tumor volume quadruples, then the dada is not complete, and the missing data arises. At the same time, for the effect of the drug, the tumor may grow more and more small, and when it reaches a cartain degree(for example < 0.01cm~3), as a result of experimental condition limit, we couldn't get the accurate tumor volume, there arises another kind of incomplete data: truncation data. How to deal with the kind of incomplete data is relatingto the clinical experimental appraisal directly. So we discuss how to analysis the two kind incomplete data with EM algorithm and the maximun likehood estimate and propose the test method correspondingly.