A study of the nature and characteristics of networks, with an application to disease spread modelling.

摘要

The natural occurrence or intentional release of highly contagious agents of livestock and poultry disease could have serious consequences for any country's agricultural economy. Successful control and management of animal disease outbreaks requires that adequate response strategies be developed beforehand. Disease spread simulation models are one method used to evaluate strategies for the control of disease spread. These models enable emergency preparedness personnel to explore many different scenarios to determine the effects of control measures such as vaccination, and predict the likely size, duration and cost of outbreaks. When researchers study these networks through modelling they assume connections between individuals are random. Recent publications have introduced new network connection structures which approximate relationships of natural networks better than random networks. This thesis builds on these results, presenting a study of disease spread in livestock using different network architectures, and showing the effect of connection structure on simulation results with strategies to halt the spread of disease. The results will help researchers to better understand the spread and containment of disease, bringing epidemics under control faster, saving the lives of animals and limiting financial hardship for farmers. The experiments are explained in the context of disease spread in animals, but the results can also be understood and applied more generally as simply nodes and links, with applications outside of animal disease spread modelling.