Integrating Socio-Demographic Attributes into Network Models of Infectious Disease Spread and Voluntary Vaccination

Infectious disease transmission is primarily driven by close-range proximity interactions. These interactions are shaped by complex human behaviour and diverse social structures. Households are known as important settings for disease propagation due to frequent and prolonged contact among household members. However, there is a notable lack of detailed, contextspecific empirical data on contact patterns within diverse household structures, especially in sub-Saharan Africa, a region with the highest burden of infectious diseases. Moreover, most of the existing epidemic models often simplify social mixing by assuming homogeneity (i.e., every individual is equally likely to be in contact with any other), neglecting the influence of socio-demographic and structural variables. At the same time, the success of preventative measures like vaccination often relies on voluntary uptake decisions, which are influenced by individual risk perceptions and the structure of social interactions, yet traditional game-theoretic models of voluntary vaccination in most cases assume a homogeneous population. Such assumptions ignore and fail to capture the significant impact that network structure and heterogeneity can have on individual decision-making and the resulting population vaccination coverage. The thesis goes beyond this common approach (mean-field or homogeneous mixing) in two key ways: integrating socio-demographic factors with social contact patterns and using a game-theoretic network approach to model vaccination behaviour. Firstly, we analyse high-resolution wearable proximity sensor data collected in rural and urban South African households to quantify and characterise within-household contact patterns. We demonstrate that household type and the gender of the household head significantly influence interaction patterns, in particular individual household roles such as child caregiving, and quantify how these heterogeneities impact the potential for disease transmission, reflected in the basic reproduction number (R0). Secondly, we develop and apply a network-based evolutionary game-theoretical model, building on the SIR model, to simulate voluntary vaccination uptake. This model incorporates heterogeneous network structures to determine the evolutionarily stable vaccination level based on the relative cost of vaccination (ratio of vaccine morbidity risk to infection morbidity risk). Our findings provide novel empirical evidence on heterogeneous household mixing in a high-burden setting, showing that structural attributes beyond age are relevant determinants of interaction patterns and disease spreading potential. The game-theoretical analysis shows that network heterogeneity significantly influences voluntary vaccination dynamics; while highly heterogeneous networks may achieve high coverage at low vaccination costs, their uptake levels decrease more sharply as costs rise compared to more homogeneous networks. By combining real-world evidence on how sociodemographic attributes shape contact patterns with network-based models of vaccination behaviour, this research offers a more realistic understanding of epidemic processes. The result has a clear implication for public health practice. Interventions must be customised according to the sociodemographic attributes of the population and the ways people perceive and act on the risk of specific interventions, such as vaccination, rather than assuming homogeneous mixing behaviour