Activity Number:
|
562
- Regression Methods for Neuroimaging Data
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 11, 2022 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistics in Imaging
|
Abstract #323330
|
|
Title:
|
Optimal Ridge Penalization in High-Dimensional Testing
|
Author(s):
|
Iris Ivy Gauran* and Zhaoxia Yu and Hernando Ombao
|
Companies:
|
King Abdullah University of Science and Technology and University of California Irvine and King Abdullah University of Science and Technology
|
Keywords:
|
EEG coherence;
Ridge Penalization;
High Dimensional Testing;
Imaging Genetics;
Generalized Cross Validation;
Genome Wide Association Studies
|
Abstract:
|
Imaging genetics involves a deluge of functional and structural data on brain-relevant genetic polymorphisms to provide insights into the neural architecture through which psychopathology may emerge. However, the high-dimensional nature of the data poses a problem on the proliferation of false discoveries. Existing methods suggest a strong regularization to prevent overfitting where the choice of tuning parameters is mostly applicable to low-dimensional data. We propose a class of methods for choosing the optimal ridge parameter and incorporate this in the adaptive Mantel test for evaluating the association of two high-dimensional sets of features. We demonstrate that the optimal value of the ridge penalty can be negative. Simulation studies show that our procedures can control the proportion of false positives with superior empirical power when applied to penalized high-dimensional testing scenarios. The application of our work to an imaging genetics study and a biological study will be presented.
|
Authors who are presenting talks have a * after their name.