Testing a Global Null Hypothesis Using Cross Validated Area Under the ROC Curve

• Activity Number: 443 - Latent Variables, Causal Inference, Machine Learning and Other Topics in Mental Health Statistics
• Type: Contributed
• Date/Time: Thursday, August 12, 2021 : 4:00 PM to 5:50 PM
• Sponsor: WNAR
• Author(s): Sunwoo Han* and Youyi Fong and Ying Huang
• Companies: Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center
• Keywords: Hypothesis test; Permutation test; Machine learning; AUC; Two-phase sampling design; Stacking

Abstract

Testing a global association between a large set of biomarker measurements and a binary outcome of disease is an important task in biomedical studies. In this talk, we propose a prediction performance-based hypothesis testing approach that combines a permutation test using cross validated area under the receiver operating characteristic curve and machine learning methods for detecting the global association. In particular, we focus on introducing our proposed testing approach under two-phase sampling designs, which are common in the biomedical studies, by dealing with an issue arose from inverse sampling probability weights. We further study the size and power of the proposed test in linear and nonlinear scenario and develop the use of stacking method to achieve competitive power in both scenarios. Illustration with an immunologic marker dataset from an HIV vaccine efficacy trial is also provided.