Wayne Getz, University of California, Berkeley

Abstract

Dynamic SEIR (Susceptible, Exposed, Infectious, Removed) compartmental models provide a tool for predicting the size and duration of unfettered outbreaks, and exploring the impacts of interventions on epidemics (e.g., isolation of infectious individuals, vaccination of susceptible individuals). The performance of these models as adequate representations of reality (Getz et al, 2018) depends on the extent to which the assumptions that: i) the same disease progression processes and rates apply equally to all individuals in the population, and ii) the population itself is well-mixed. Variability in individual's behavior, exposure history, immunological and coinfection states, requires a vast amount of data compared with a single time series of incidence data that is used to fit SEIR models to specific epidemics. Incidence data, however, typically have a spatial element to them because these data often come tagged with a location where they were collected (e.g. clinic, town, or city). Here we explore ways to make use of incidence data aggregated at levels of spatial resolution where the assumption that the outbreak consists of a network of process and spatially homogeneous subpopulations to which SEIVD (naturally Vaccinated, Dead) models can be locally applied. Our continuous and discrete-time, deterministic and stochastic versions of these models are constructed with the Numerus Model Builder platform (Getz and Dougherty, 2018; Getz et al, in press) , which permits direct extension to metapopulation versions and one-click generation of web browser implementations. We apply these models to incidence data from the 2014 West African Ebola outbreak, focusing on a comparison of the disease process parameters that locally apply and the degree to which it may be reasonable to apply average parameters to making predications at a country level.

References:

Getz, W. M., C. R. Marshall, C. J. Carlson1, L. Giuggioli, S. J. Ryan, C. Boettiger, S. D. Chamberlain,L. Larsen, P. D’Odorico, D. O’Sullivan, S. S. Romañach, 2018a. Making ecosystem models adequate. Ecology Letters, 21 (2): 153-166.

Getz, W. M. and E. J. Dougherty. Discrete Stochastic Analogs of Erlang Epidemic Models, 2018. Journal of Biological Dynamics,12 (1), 16-38

Getz, W. M., R. Salter, O. Muellerklein, H. S. Yoon, K. Tallam, in press. An Epidemic Modeling Primer and Nova Model Builder Implementation. Epidemics.