Inference for a spatio-temporal model with partial spatial data: African horse sickness virus in Morocco example

An example randomly generated region-level spatial distribution and code associated with Emma L. Fairbanks, Matthew Baylis, Janet M. Daly, Michael J. Tildesley. Inference for a spatio-temporal model with partial spatial data: African horse sickness virus in Morocco, Epidemics, 2022, 100566, ISSN 1755-4365, https://doi.org/10.1016/j.epidem.2022.100566. (https://www.sciencedirect.com/science/article/pii/S1755436522000202) Abstract: African horse sickness virus (AHSV) is a vector-borne virus spread by midges (Culicoides spp.). The virus causes African horse sickness (AHS) disease in some species of equid. AHS is endemic in parts of Africa, previously emerged in Europe and in 2020 caused outbreaks for the first time in parts of Eastern Asia. Here we analyse a unique historic dataset from the 1989-1991 emergence of AHS in Morocco in a naïve population of equids. Sequential Monte Carlo and Markov chain Monte Carlo techniques are used to estimate parameters for a spatial–temporal model using a transmission kernel. These parameters allow us to observe how the transmissiblity of AHSV changes according to the distance between premises. We observe how the spatial specificity of the dataset giving the locations of premises on which any infected equids were reported affects parameter estimates. Estimations of transmissiblity were similar at the scales of village (location to the nearest 1.3 km) and region (median area 99 km2), but not province (median area 3000 km2). This data-driven result could help inform decisions by policy makers on collecting data during future equine disease outbreaks, as well as policies for AHS control. Keywords: Vector-borne disease; Spatio-temporal model; Bayesian inference