Principal Investigator: Dr Yelena Tarasenko
Georgia Southern University, Georgia, USATags: 44084, Bayesian, cancer, cardiovascular disease, effect modification, gene-environment interaction, physical activity
Collaborator: Dr Daniel Linder, Augusta University, Georgia, USA
Many studies have documented main or independent effects of various demographic, lifestyle, genomic, clinical, and psychosocial factors on cancer and cardiovascular outcomes and engagement in physical activity. However, we know little about interaction effects; i.e., the combined effects of multiple exposures or risk factors and effects of one exposure within strata of another. An improved understanding of these effects is essential for uncovering synergistic or differential relationships and preventing development of ineffective or mistargeted interventions. Hence, our goal is to identify interaction and effect modification within the physical activity and cancer control framework.
Because interactions are rare and usually unknown, identifying them from data has proven to be challenging, in terms of statistical inference and the requisite computation in potentially high-dimensional spaces. Our first aim is to develop novel Bayesian approach which will offer several advantages over an already scant number of approaches to detecting interactions. Our second aim is to apply the developed methodology to the UK Biobank – “the world’s largest objective physical activity dataset now available,” which “redefined what is possible in the field of physical activity epidemiology,” and paved the way to “robustly examine associations and interactions between activity, diseases, environmental factors and genetics” (UK Biobank, 2/2/2017).
The project will be executed in three years.