| Brief Introduction to Speaker |
Abstract
We propose and study a semiparametric model for the analysis of sub-groups in the clinical trial study, in which one subgroup is particular effective for the treatment, and another is not, to be classified according to the subjects' covariates. The model is formulated as a geometrical mean of a parametric and a nonparametric components. The parametric component represents our knowledge about the model and the nonparametric part represents the uncertainty. The profile likelihood method is used to infer the model parameters, the profile likelihood ratio and score statistics are used to test the existence of subgroups. As the subgroup with particular effect for the treatment is of more clinical importance, if the existence of subgroup is confirmed, we use Neyman-Pearson rule is used to classify each subjects to one of the subgroups, so that the misclassification error for the treatment favored group is under control by pre-specified criterion while that for the other subgroup is minimized.
|
