Predictive risk modelling and mapping of Murray Valley Encephalitis virus and Dengue virus in Western Australia

Murray Valley Encephalitis virus (MVEV) and Dengue virus (DENV) are two medically important arboviruses that cause significant morbidity and occasional mortality in Australia. To date, there is no direct antiviral treatment for these two diseases and no effective vaccine exists to prevent their spread in the community. Surveillance, response, and public education remain the most effective tools for health authorities to monitor and control them.

Risk modelling and mapping are two major components of disease surveillance. They aim to capture the major factors that influence the emergence of disease into a predictive framework, which is then used to identify potential hot-spots of disease occurrence. This thesis contributes to those aims by developing two risk models for MVEV and DENV, and applying them towards identifying regions in Western Australia that are potentially at risk of experiencing outbreaks of these viruses. The models were constructed using Bayesian-Belief Networks (BBNs): graphical networks that represent conditional probability relationships between random variables (which in this context are the factors affecting MVEV and DENV risks). Bayesian Belief Networks are particularly suited to modelling large systems with multiple interacting components, such as the ecosystems in which MVEV and DENV survive and propagate.

Comparisons of the risk maps based on published data and historical records show that they are able to predict high risk locations with some accuracy, increasing our confidence in their overall validity. The risk models and maps developed in this thesis could add to the range of surveillance measures available to combat MVEV and DENV in Western Australia.