A&S CREATES Presents: Decoding social and environmental drivers of accelerated biological aging with machine learning
Peter Song, Professor of Biostatistics at University of Michigan
Do environmental toxicants accelerate biological aging? Conversely, could accelerated biological aging contribute to an increased risk of preterm birth (PTB) compared with full-term birth (FTB)? Such questions are increasingly important in environmental health research for understanding the interplay among social and environmental exposures, biological aging, and health outcomes. DNA methylation (DNAm)-based epigenetic age provides an important measure of biological aging, but conventional mean-based approaches may overlook substantial heterogeneity among individuals. Quantile-based epigenetic age clocks offer a more sensitive approach to capturing such heterogeneity.
In this talk, we present machine learning approaches for investigating social and environmental drivers of quantile-specific aging acceleration. In the ELEMENT cohort of adolescents aged 8-18 years, we identify significant associations between polycyclic aromatic hydrocarbon (PAH) exposures and quantile aging acceleration that are undetected by conventional mean-based methods. Similarly, in the BIBB cohort, Black pregnant women with PTB and FTB exhibit significant differences in quantile aging acceleration, whereas mean aging acceleration shows no significant difference. These findings highlight the potential of quantile-based epigenetic clocks to uncover important exposure-aging-health relationships masked by mean-based analyses.
Learn more about A&S CREATES here.