Modeling West Nile Virus transmission in birds and humans: Advantages of using a cellular automata approach
Baki Cissé,David R. Lapen,K. Chalvet-Monfray,Nicholas H. Ogden,Antoinette Ludwig
DOI: https://doi.org/10.1016/j.idm.2024.01.002
2024-02-04
Infectious Disease Modelling
Abstract:In Canada, the periodic circulation of West Nile Virus (WNV) is difficult to predict and, beyond climatic factors, appears to be related to the migratory movements of infected birds from the southern United States. This hypothesis has not yet been explored in a spatially distributed model. The main objective of this work was to develop a spatially explicit dynamic model for the transmission of WNV in Canada, that allows us to explore non-climate related hypotheses associated with WNV transmission. A Cellular Automata (CA) approach for multiple hosts (birds and humans) is used for a test region in eastern Ontario, Canada. The tool is designed to explore the role of host and vector spatial heterogeneity, host migration, and vector feeding preferences. We developed a spatialized compartmental SEIRDS-SEI model for WNV transmission with a study region divided into 4 km2 rectangular cells. We used 2010–2021 bird data from the eBird project and 2010–2019 mosquito data collected by Ontario Public Health to mimic bird and mosquito seasonal variation. We considered heterogeneous bird densities (high and low suitability areas) and homogeneous mosquito and human densities. In high suitability areas for birds, we identified 5 entry points for WNV-infected birds. We compared our simulations with pools of WNV-infected field collected mosquitoes. Simulations and sensitivity analyses were performed using MATLAB software. The results showed good correspondence between simulated and observed epidemics, supporting the validity of our model assumptions and calibration. Sensitivity analysis showed that a 5% increase or decrease in each parameter of our model except for the biting rate of bird by mosquito ( c(B,M)) and mosquito natural mortality rate ( dM ), had a very limited effect on the total number of cases (newly infected birds and humans), prevalence peak, or date of occurrence. We demonstrate the utility of the CA approach for studying WNV transmission in a heterogeneous landscape with multiple hosts.
infectious diseases,mathematical & computational biology