Bi-linear matrix-variate analyses, integrative hypothesis tests, and case-control studies

摘要: We pursue a threefold purpose in this paper. First, we suggest Kullback-Leibler formulation for developing statistics and making discriminative projection case-control studies, based on which existing typical methods are revisited then further extended to matrix-variate counterparts. Second, propose bi-linear matrix form, multivariate analysis logistic, Cox, linear mixed regression into their Third, systematically address the necessity, feasibility, methodology of integrative hypothesis tests (IHT) from complementarity model-based test boundary-based (BBT) data (D)-space, (S)-space, probability (P)-space. elaborate four IHT components (modelling, comparison, classification, assurance) summarise types D-space. Then, extend efforts BBTs S-space. Particularly, classic univariate one-tail z-test ones, is applied sample-pairing delta (SPD) detecting collective inclining dominance. Also, SPD that extends test. Moreover, bi-test null also about inference reliability due space complexity, including development Fisher combination. Finally, possible applications gene expression biomarkers exome-sequencing-based joint single-nucleotide variant (SNV) detection.