报告题目(Title): A Statistical Framework for Temporal Multi-omics Data

报告时间(Time)： 2026年7月29日12:00-13:00

报告人（Speaker）：李赫博士

Department of Mathematics, Radboud University, Netherlands

报告地点：腾讯会议ID：161 391 741

报告摘要(Abstract): Longitudinal studies comprising high-dimensional molecular measurements provide unique opportunities to model biological signals over time. Current methods cannot simultaneously reduce dimensionality, perform inference on time effects, and integrate multi-omics datasets. We propose a novel unified framework that combines probabilistic principal component analysis with linear mixed models to address these challenges. The framework consists of dynamic principal component analysis (dPCA) for a single dataset, and dynamic partial least squares (dPLS) for the joint analysis of two correlated datasets. In this talk, we will present the simulation studies that accurately recover parameters across four settings. We will also illustrate the novel insights by applying the framework to the TwinsUK cohort.