Online Program Home
  My Program

Abstract Details

Activity Number: 465 - Biometrics and High-Dimensional Data
Type: Contributed
Date/Time: Wednesday, August 2, 2017 : 8:30 AM to 10:20 AM
Sponsor: Biometrics Section
Abstract #324026
Title: Test for a Change Point in the Linear Combination of Risk Predictors
Author(s): Ju Hee Cho* and Ying Huang
Companies: Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center
Keywords: change point model ; heterogeneous subgroups ; maximum of score statistics ; marker combination ; multivariate biomarkers
Abstract:

In change point regression program, methods have been developed to test the threshold effect based on a univariate covariate, through maximum of score or likelihood-ratio statistics across various thresholds. We extended these approaches to the case of multivariate covariates. We are interested in testing for the existence of heterogeneous subgroups characterized by the change point in a linear combination of multivariate covariates. The proposed method is computationally more efficient than the previous method for dealing with multivariate cases but is equally or more powerful. Also, compared to an alternative two-stage method that first estimates the marker combination and then performs a hypothesis test using the estimated marker combination, the proposed method does not require the estimation of marker combination before testing, and is thus simpler but achieves better performances. Theoretical and numerical analyses corroborate these arguments. We illustrate application of the proposed method using data example from HIV vaccine research where it is of interest to identify subgroups with heterogeneous risk of HIV infection based on vaccine-induced immune biomarkers.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association