Simulating infectious diseases using network and individual-based models

The availability of data on the movement of cattle within the United Kingdom has inspired a substantial number of models of infectious disease dynamics; a proportion of these models use techniques based upon network analysis, and have been deployed to investigate economically significant diseases such as bovine tuberculosis and foot and mouth disease. We model the movements of UK cattle as a directed network, with animal holdings as nodes, and movements of animals as edges. Within this framework, a range of models may be considered, from a simple static network based upon all movements (sampled from a fixed number of days), to a more realistic and more complex dynamic model where the network changes every day. Whilst the former may be analysed with a far broader range of techniques, we demonstrate using stochastic simulation that they fail to adequately capture the aspects of network structure that drive the dynamics of disease processes. As a contrast to these network models of disease transmission, we also simulate the spread of disease, but modelling the dynamics at the level of individual cattle, rather than farms -hence providing a more detailed picture of disease spread. Comparing the output of these individualbased models to the output of network-based models is challenging (since the latter do not account for the disease status of individuals, and the former do not neatly describe the disease status of holdings); we discuss some approaches to making this commparison, and show how it may aid the interpretation of network-based models of animal disease