CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: A shrinkage likelihood ratio test for high-dimensional subgroup analysis with a logistic-normal mixture model Authors:  Shota Takeishi - University of Tokyo (Japan) [presenting]
Abstract: The focus is on testing the existence of a subgroup with an enhanced treatment effect under the setting where the subgroup membership is potentially characterized by high-dimensional covariates. Using a logistic-normal mixture model, we propose a shrinkage likelihood ratio test built on a modified likelihood function that shrinks high-dimensional unidentified parameters towards zero when there exists no subgroup. This shrinkage helps handle the irregularity of the testing problem in the logistic-normal mixture model. It enables us to derive a tractable chi-squired-type asymptotic null distribution even under the high-dimensional regime.