2015/06/01 by Chris Jewell, Jewell, Chris, Brown, Richard +1
Agricultural and Biological Sciences · #Vector-Borne Animal Diseases #Insect and Pesticide Research #Animal Disease Management and Epidemiology
paper · pdf · doi:10.48550/arxiv.1506.00602
Predicting the spread of vector-borne diseases in response to incursions\nrequires knowledge of both host and vector demographics in advance of an\noutbreak. Whereas host population data is typically available, for novel\ndisease introductions there is a high chance of the pathogen utilising a vector\nfor which data is unavailable. This presents a barrier to estimating the\nparameters of dynamical models representing host-vector-pathogen interaction,\nand hence limits their ability to provide quantitative risk forecasts. The\nTheileria orientalis (Ikeda) outbreak in New Zealand cattle demonstrates this\nproblem: even though the vector has received extensive laboratory study, a high\ndegree of uncertainty persists over its national demographic distribution.\nAddressing this, we develop a Bayesian data assimilation approach whereby\nindirect observations of vector activity inform a seasonal spatio-temporal risk\nsurface within a stochastic epidemic model. We provide quantitative predictions\nfor the future spread of the epidemic, quantifying uncertainty in the model\nparameters, case infection times, and the disease status of undetected\ninfections. Importantly, we demonstrate how our model learns sequentially as\nthe epidemic unfolds, and provides evidence for changing epidemic dynamics\nthrough time. Our approach therefore provides a significant advance in rapid\ndecision support for novel vector-borne disease outbreaks.\n