When: Friday, March 13, 3:00 PM
Where: Tyler 055

Abstract
Biological aging emerges from dynamic interactions between molecular systems and longitudinal physiological change, yet most aging models rely on single-omics or cross-sectional data and often fail to generalize across heterogeneous cohorts. We present a unified framework for cross-modal and cross-cohort learning in multi-omics integration to enable robust, interpretable aging models. First, we introduce ONDSA, a false discovery rate (FDR)-controlled network testing framework that compares multi-omics Gaussian graphical models to identify shared and differential pathways across biological groups, filling a key gap in multi-group precision matrix inference with rigorous error control. Second, we develop MGRFusionNet, a multi-modal graph-recurrent fusion architecture that integrates baseline molecular networks with longitudinal phenotypic trajectories to capture nonlinear, time-dependent risk dynamics. In the Framingham Heart Study, MGRFusionNet outperforms unimodal and naive fusion baselines for mortality risk prediction while enabling interpretable attribution of cross-modal interactions. Finally, we develop Trans-GCR, a cross-cohort transfer learning approach that borrows strength from large biobanks for smaller, clinically rich target cohorts. Trans-GCR explicitly models domain shift in network-structured data and offers theoretical guarantees with computational efficiency in high-dimensional settings, supporting stable learning in small target cohorts. Together, these methods provide a scalable, interpretable foundation for next-generation multi-omics aging research.

Bio
Jiachen Chen is a PhD candidate in Biostatistics at the Boston University School of Public Health. Her research focuses on developing statistical and machine learning methods for integrating high-dimensional multi-omics data, with an emphasis on interpretable and generalizable models for population-based studies. Her work is motivated by aging-related applications, including frailty, cognitive decline, and dementia. She is an NIA F99/K00 Fellow, supported by the National Institute on Aging and a Boston University Center for Health Data Science Pilot Award. Her research has been published in journals including Briefings in Bioinformatics, Statistics in Medicine, Nature, JAMA Psychiatry, Alzheimer’s & Dementia, and Aging Cell.