Spatiotemporal Fidelity of a Metapopulational Model Evaluated on the COVID-19 Pandemic in Sweden

Abstract: A modified SEIR compartmental model is constructed for COVID-19 in a metapopulational setting with fine-scaled population data. The model is stochastically simulated with GLEAMviz that provides realistic short and long distance mobility based on real-world data. A retrospective evaluation in both the temporal and spatial dimensions is conducted with over one year of collected data from the COVID-19 pandemic in Sweden. We find that to reproduce the multimodal behavior a seasonal scaling factor of 0.4 is necessary, which is introduced to the model by scaling R0 with corresponding sinusoidal. For the spatial dimension we divide Sweden into a southern, middle and northern region and the model is able to capture the dynamics in all regions. Additionally, we introduce compartmental models in a constructive manner and motivate metapopulational models, including how commuting is integrated by a force of infection. The next generation method for calculating the basic reproductive number for arbitrary compartmental models is also presented.