A spatio-temporal network epidemiological agent-based modelling approach for mapping risk and exposure of viruses

Abstract

With the globalization of the world, the ability to contain viruses becomes more important. Modeling the spread of a virus is a great tool for understanding how to prevent future epidemics from happening. This thesis continues on an existing model of the COVID-19 pandemic in The Netherlands, which is an epidemiological agent-based model on a spatio-temporal network. The model is adjusted and improved in order to create risk and exposure maps of The Netherlands at a municipality level, per demographic group. The risk maps show the total infections after three weeks after infection of five agents in a certain municipality, while the exposure maps show where the exposures to the virus happened. In order to obtain the desired risk maps, the original model is extended with extra epidemiological parameters and features that allow for infections to be initialised with high precision. The resulting risk maps show that the risk of a municipality depends mostly on the average mobility per agent, as well as the location of the municipality in the underlying complex network of mobility, demography and behavior. As expected, the Randstad is clearly the most risky area in The Netherlands for the high mobility demographic groups. The major municipalities like Amsterdam, Rotterdam and Utrecht do not stand out as higher risk compared to other nearby municipalities, however they are still responsible for the vast majority of exposures to the virus. The results show possibilities for future research into new intervention measures, to discover the efficiency of significantly reducing mobility to these exposure locations, no matter where the infections originated from.