|
Abstract:
|
Subgroup identification in precision medicine is to identify patient subgroups with significant differential treatment effect based on patient's characteristics such as genomic biomarkers, disease history, etc., so that the benefit of treatment can be optimized. Although currently there are many available data mining algorithms that can be applied to subgroup identification, most of them are focusing on class prediction and accuracy instead of finding the patterns that describe our interested responder subpopulation. In this study, we use subgroup discovery algorithms to develop predictive signature to find the largest subgroup of responders that share similar characteristics. After applying subgroup discovery algorithms to sample data, only those patients identified by the predictive signature are kept as our primary subgroup and the remaining patients are ignored. A simulation study is conducted to compare the performance of subgroup discovery algorithms with that of some other recursive partitioning algorithms regarding the power as well as the type I error.
|
