Abstract:Modern high-throughput biomedical devices routinely produce data on a large scale, and the analysis of high-dimensional datasets has become commonplace in biomedical studies. However, given thousands or tens of thousands of measured variables in these datasets, extracting meaningful features poses a challenge. In this article, we propose a procedure to evaluate the strength of the associations between a nominal (categorical) response variable and multiple features simultaneously. Specifically, we propose a framework of large-scale multiple testing under arbitrary correlation dependency among test statistics. First, marginal multinomial regressions are performed for each feature individually. Second, we use an approach of multiple marginal models for each baseline-category pair to establish asymptotic joint normality of the stacked vector of the marginal multinomial regression coefficients. Third, we estimate the (limiting) covariance matrix between the estimated coefficients from all marginal models. Finally, our approach approximates the realized false discovery proportion of a thresholding procedure for the marginal p-values, for each baseline-category pair. The proposed approach offers a sensible trade-off between the expected numbers of true and false rejections. Furthermore, we demonstrate a practical application of the method on hyperspectral imaging data. This dataset is obtained by a matrix-assisted laser desorption/ionization (MALDI) instrument. MALDI demonstrates tremendous potential for clinical diagnosis, particularly for cancer research. In our application, the nominal response categories represent cancer subtypes.

Two-Sample Covariance Matrix Testing and Support Recovery in High-Dimensional and Sparse Settings

Testing and Support Recovery of Multiple High-Dimensional Covariance Matrices with False Discovery Rate Control

Double verification for two‐sample covariance matrices test

Two Sample Tests for High Dimensional Covariance Matrices

High-dimensional Two-Sample Mean Vectors Test and Support Recovery with Factor Adjustment.

HIGH-DIMENSIONAL TWO-SAMPLE COVARIANCE MATRIX TESTING VIA SUPER-DIAGONALS

More Powerful Tests for Sparse High-Dimensional Covariances Matrices

Two sample test for covariance matrices in ultra-high dimension

Multi-level Thresholding Test for High Dimensional Covariance Matrices

Testing and Signal Identification for Two-Sample High-Dimensional Covariances Via Multi-Level Thresholding

On testing the equality of high dimensional mean vectors with unequal covariance matrices

Testing Super-Diagonal Structure in High Dimensional Covariance Matrices

Bayesian Optimal Two-sample Tests in High-dimension

SHARP OPTIMALITY FOR HIGH-DIMENSIONAL COVARIANCE TESTING UNDER SPARSE SIGNALS

Use of Random Integration to Test Equality of High Dimensional Covariance Matrices

HYPOTHESIS TESTING ON LINEAR STRUCTURES OF HIGH DIMENSIONAL COVARIANCE MATRIX.

Multiple multi-sample testing under arbitrary covariance dependency

Testing and support recovery of correlation structures for matrix-valued observations with an application to stock market data

Two-Sample Test of High Dimensional Means under Dependence

Tests for High-Dimensional Covariance Matrices Using Random Matrix Projection

A New Statistic for Testing Covariance Equality in High-Dimensional Gaussian Low-Rank Models