Local-level spatiotemporal dynamics of COVID-19 transmission in the Greater Seoul Area, Korea: a view from a Bayesian perspective.

Objectives: The purpose of this study was to enhance the understanding of the local-level spatiotemporal dynamics of COVID-19 transmission in the Greater Seoul Area (GSA), Korea, after its initial outbreak in January 2020.
Methods: Using the weekly aggregates of coronavirus disease 2019 (COVID-19) cases of 77 municipalities in the GSA, we examined the relative risks of COVID-19 infection across local districts over 50 consecutive weeks in 2020. To this end, we employed a spatiotemporal generalized linear mixed model under the hierarchical Bayesian framework. This allowed us to empirically examine the random effects of spatial alignments, temporal autocorrelation, and spatiotemporal interaction, along with fixed effects. Specifically, we utilized the conditional autoregressive and the weakly informative penalized complexity priors for hyperparameters of the random effects.
Results: Spatiotemporal interaction dominated the overall variability of random influences, followed by spatial correlation, whereas the temporal correlation appeared to be small. Considering these findings, we present dynamic changes in the spread of COVID-19 across local municipalities in the GSA as well as regions at elevated risk for further policy intervention.
Conclusions: The outcomes of this study can contribute to advancing our understanding of the local-level COVID-19 spread dynamics within densely populated regions in Korea throughout 2020 from a different perspective, and will contribute to the development of regional safety planning against infectious diseases.