Abstract for: A Method for Modelling and Calibrating Disaggregated Diffusion Models

Complex dynamic problems, such as infectious disease spread, have at their core diffusion processes that are driven by reinforcing feedback loops. System dynamics approaches tend to view such diffusion problems at an aggregate level, based on the assumption of random mixing within the population. However, in the public health area, assortative (within-type) mixing is a recognised empirical phenomenon, and therefore simulation models must disaggregate across key cohorts in order to maximize engagement with policy makers, and provide more robust and accurate models of disease spread. This paper integrates key ideas from modern infectious disease modeling approaches into a system dynamics context, and presents key formulations to allow for the disaggregation of SD diffusion models. It also shows how case data can be aligned with structural SD models, thereby allowing the model to be calibrated and fit to historical data. The approach is validated using an SEIR model, and based on a case study of the 1957 flu outbreak in the United Kingdom.