Previous research has highlighted the significant role social networks play in the spread of non-communicable chronic diseases. In our research, we seek to explore the impact of these networks in more detail and gain insight into the mechanisms that drive this. We use obesity as a case study. To achieve this, we develop a generalisable hybrid simulation and optimisation approach aimed at gaining qualitative and quantitative insights into the effect of social networks on the spread of obesity. Our simulation model has two components. Firstly, an agent-based component mimics the dynamic structure of the social network within which individuals are situated. Secondly, a system dynamics component replicates the relevant behaviours of those individuals. The parameters from the combined model are refined and optimised using longitudinal data from the United Kingdom. The simulation produces projections of Body Mass Index broken down by different age groups and gender over a 10-year period. These projections are used to explore a range of scenarios in a computational study designed to address our research aims. The study reveals that, for the youngest population sub-groups, the network acts to magnify the impact of external and social factors on changes in obesity, whereas, for older sub-groups, the network mitigates the impact of these factors. The magnitude of that impact is inversely correlated with age. Our approach can be used by public health decision makers as well as managers in adult weight management services to enhance initiatives and strategies intended to reduce obesity. Our approach is generalisable to understand the impact of social networks on similar non-communicable diseases.