Building research workforce expertise in high dimensional data analysis and communication to promote rigorous research

About This Grant

Summary The next generation of biostatisticians and epidemiologists will routinely be considering complex biomarker data. Examples include multiplex cytokine data, microbiome data, -omics and require analyses of high-dimensional, often correlated data, where the interest is in describing pathways or communities, and their influence on health. Many of these datasets will be collected in longitudinal studies as well, adding to the analytic challenge and requiring advanced training in analytical methods for describing systems rather than single exposures. These complex exposures are often best examined using tools developed in fields such as quantitative community ecology and computational biology, which are outside the realm of traditional epidemiologic and biostatistical training. This cross-disciplinary training program will teach trainees the tools for handling high-dimensional exposures and outcomes including ordination methods, clustering techniques, graph-theoretic approaches, data reduction strategies and network analysis. A focus on data visualization techniques and data communication will position trainees to both analyze increasingly ubiquitous biomarker data and effectively communicate inferences to collaborators, funders and the public. Trainees will also gain skills in reproducible research to ensure rigor and transparency. There is a clear imperative to ensure that the evidence base that could lead to or away from promising new treatments or unlock the door to mechanistic insights remains unpolluted by poorly executed science. We need highly trained analysts, who understand the tools and the assumptions, to handle the complex data that are becoming increasingly pervasive. This training grant will help fill a clear gap in our analytic workforce.