Dynamic contagion spread modelling over emergent spatio-temporal contact networks
University of New Brunswick
Analysis, prediction and decision support for contagion spread and its associated risk with Ordinary Differential Equation models proved to be challenging in terms of incorporating the underlying heterogeneity of emergent spatio-temporal contact networks of the moving objects. These models reduce the granularity and scope of analysis of possible pharmacological and non-pharmacological intervention measures. To address these limitations, we propose to aggregate population level dynamics by modelling individual-level interactions. To that end, first we introduce an individual level agent-based stochastic contagion simulation modelling framework as a possible solution for adapting to rapidly changing parameters of a contagion including SARS-CoV-2. Second, we propose a spatio-temporal index to extract contact networks from distributed real-world mobility data stores. The framework has been implemented declaratively in R and functionally in Julia and the spatio-temporal index in Python. We conduct thorough experimental evaluation to show the viability and applicability of the proposed approaches.