EAGER: Collaborative Research: Rapid Production of Geospatial Network Inputs for Spatially Explicit Epidemiologic Modeling of COVID-19 in the USA

Lead PI: Dr. Christopher Small

Unit Affiliation: Marine and Polar Geophysics, Lamont-Doherty Earth Observatory (LDEO)

June 2020 - May 2021
Inactive
North America ; United States ; New York ; Los Angeles, CA
Project Type: Research

DESCRIPTION: Dynamical computer models of disease transmission are used to understand and predict how infectious diseases spread through host populations. Maps of population distribution, mobility, and travel corridors are critical components of many of these models. However, accurately determining the spatial distribution of people is difficult because most sources of data (e.g., census) indicate only approximately where people reside, rather than where they work and go. Census data, in particular, are also aggregated in a way that provides fine spatial detail only in densely populated urban areas. In suburban and rural areas, census maps provide only the total number of people living in each census unit (e.g., a U.S. county), but do not show where people live and work. This research will fuse detailed satellite images of night light emitted from cities, towns and travel corridors with census counts and mobility data to produce more detailed population maps for epidemiologists to use to more accurately simulate the transmission of communicable diseases like COVID-19. The proposed collaboration will bring together expertise from geospatial dynamics and remote sensing with disease ecology and epidemiology to produce boundary spanning science with potential to advance both fields. Further, the proposed project will support two early career scientists as well as undergraduate student involvement in research.

When air and vehicle travel are significantly reduced, the accuracy and detail of population movement and spatial connectedness assumes greater importance for modeling epidemic spread. Spatial networks derived from co-analysis of geospatial data (settlement and infrastructure density from remotely sensed night light and population density from census enumerations) can provide more accurate spatial domains than the administrative units (e.g., counties) used to aggregate and analyze health data. In addition, the structure and connectivity of these spatial networks can be used to quantify fundamental parameters of network structure that influence disease spread. This research will develop a progressively refined suite of network maps for use with epidemiological models. The research team, composed of geoscientists, disease ecologists and epidemiologists will develop a standardized protocol with analytic procedures and tools for production of these maps structured so as to be suited for quantitative spatiotemporal analysis of SARS-CoV-2 infections in the U.S., including detailed analyses of the New York and Los Angeles metro areas. Network flow parameters among population centers will be estimated using agent-based modeling, establishing a complete geospatial network consisting of population and mobility constraints within cities, and population fluxes among cities. Population and network flow estimates will be input directly into spatially explicit COVID-19 transmission models, and will be abstracted into boundary conditions that can streamline future epidemiological models. This RAPID award is made by the Ecology and Evolution of Infectious Disease Program in the Division of Environmental Biology, using funds from the Coronavirus Aid, Relief, and Economic Security (CARES) Act.