YSPH Biostatistics Seminar: “Knowledge Integration Quantile Regression (KIQR) in Ultra-High Dimensions"

SPEAKER: Yan Yu, PhD, Joseph S. Stern Professor of Business Analytics, Department of Operations, Business Analytics, and Information Systems, University of Cincinnati

TITLE: “Knowledge Integration Quantile Regression (KIQR) in Ultra-High Dimensions”

ABSTRACT: We propose a Knowledge Integration Quantile Regression (KIQR) framework for simultaneous variable selection and estimation in ultra-high dimensions, motivated by an application to identifying genetic risk factors for obesity from hundreds of thousands of single nucleotide polymorphisms (SNPs) in a cohort study with a limited number of participants. KIQR integrates prior information from external large-scale studies such as the GIANT consortium and UK Biobank into a penalized quantile regression framework via a novel knowledge integration mechanism that adaptively balances prior knowledge against observed data while remaining robust to potential prior misspecification. Focusing on high conditional quantiles of body mass index (BMI) most relevant to obesity risk, KIQR addresses key limitations of standard genome-wide association studies (GWAS): the stringent significance threshold, the univariate marginal mean association, and the lack of utilization of prior knowledge. We establish oracle properties and large sample properties of the penalized estimator under the non-differentiable check loss via Huber approximation, combined with a nonconvex SCAD penalty handled through local linear approximation. Simulation studies further demonstrate its effectiveness. Applied to the Framingham Heart Study, KIQR identifies three novel SNP associations: rs3798696 in TFAP2A, rs7070523 in ITIH5, and rs178260 in AIFM3, which have not previously been reported in the GWAS literature.

YSPH values inclusion and access for all participants. If you have questions about accessibility or would like to request an accommodation, please contact Charmila Fernandes at Charmila.fernandes@yale.edu. We will try to provide accommodations requested by September 29, 2026.