A trans-disciplinary approach to disease ecology: linking local to regional scales of drivers and biological processes through big data-model integration

Predicting the drivers of spread of vector-borne diseases is challenging given that temporal variability in climate and heterogeneity in geospatial variables can influence relationships among hosts, pathogens, and vectors in complex ways. Most arthropod-borne viruses (arboviruses) infect livestock, humans or wildlife hosts on local scales that can cascade through multiple dispersal and transport mechanisms to influence much larger geographic extents. The availability of long-term climate and geospatial data for numerous drivers at multiple scales provides opportunities for application to a predictive disease ecology paradigm provided these data can be synthesized and harmonized with the fine-scale, highly resolved data on host and vector responses to disease and the environment. Our goal was to develop an operational framework for predictive disease ecology using big data and trans-disciplinary scientific expertise based on spatiotemporal modeling of cross-scale interactions coupled with human and machine learning.

We illustrate our framework with questions related to drivers of patterns in the spread of vesicular stomatitis virus (VSV) that causes the most common vector-borne vesicular disease of livestock throughout the Americas. VSV spreads at variable rates and patterns from Mexico northward across the western US often following waterways in initial outbreak years (2004, 2012-2014) followed by expansion years (2005, 2015) which typically involve regions with drier climates. We sought to provide mechanistic explanations for these patterns by first creating a trans-disciplinary team of scientists and software engineers from across the US to develop a conceptual model of the host-virus-environment system. We then applied a multi-scale big data-model integration analysis framework with human and machine learning to determine the relative importance of >400 biological, climatic, and geospatial variables to patterns in occurrence of VS. Our trans-disciplinary team approach, building on technologies for big data to develop testable hypotheses about the variables driving the spread of VSV, can be applied to other emerging diseases that start at a small outbreak scale, but propagate to much larger geographic extents.