A0905
Title: Locally sparse varying coefficient mixed model with application to longitudinal microbiome differential abundance
Authors: Simon Fontaine - University of Michigan (Canada) [presenting]
Gen Li - University of Michigan Ann Arbor (United States)
Ji Zhu - University of Michigan (United States)
Abstract: Differential abundance (DA) analysis in microbiome studies has recently been used to uncover a plethora of associations between microbial composition and various health conditions. While current approaches to DA typically apply only to cross-sectional data, many studies feature a longitudinal design to understand the underlying microbial dynamics better. A novel varying coefficient mixed-effects model with local sparsity is introduced to study DA on longitudinal microbial studies. The proposed method can identify time intervals of significant group differences while accounting for temporal dependence. Specifically, a penalized kernel smoothing approach is exploited for parameter estimation, and local regression is extended to include a random effect without any requirements for the sampling design. In particular, it operates effectively regardless of whether sampling times are shared across subjects, accommodating irregular sampling or potentially missing observations. Simulation studies demonstrate the necessity of modelling dependence for precise estimation and support recovery. The method's application to a longitudinal study of mice's oral microbiome during cancer development revealed significant scientific insights that were otherwise not discernible through cross-sectional analyses.
