Dynamics of infectious diseases on mixing networks

• Main Author: Eames, K.
• Published: University of Cambridge 2004
• Subjects: 616.9
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.598725

This thesis describes the development of models designed to capture the dynamics of infections taking place within mixing networks. These models, formulated as systems of differential equations using pair-wise moment closure techniques, are investigated for a wide variety of situations and shown to be highly flexible and often very different from traditional approaches. When compared to full stochastic simulations on computer-generated networks they are remarkably accurate, giving excellent agreement both to the initial growth and the equilibrium behaviour of epidemics. The models are used to investigate a range of interventions, including targeted control measures and contact tracing. Contact tracing, which attempts to identify contacts of infected individuals, is an intrinsically network-based intervention that the pair-wise models developed here are well suited to capturing. It is shown to be a naturally targeted and powerful control that can reduce prevalence by automatically concentrating efforts on high-risk subpopulations. Network methods are particularly applicable to sexually transmitted diseases (STDs), for which networks are relatively easy to define and frequently measured. To better capture the behaviour of STDs in the general population, a monogamous network model is developed, including partnership turnover within a serially monogamous society. The influence of turnover rates is apparent, and the reduced impact of high mixing groups within such populations has ramifications for the design of control policies. The presence of partnership dynamics within monogamous networks introduces a second time-scale that allows the existence of multiple pathogen strains whereas, in a polygamous environment, only one would be able to persist. The coexistence of otherwise mutually exclusive strains has implications for disease evolution, and demonstrates the importance of population mixing patterns and behaviour.