Modelling and analysis of COVID-19 outspread at micro-levels using spatial autocorrelation: Case of eThekwini

Abstract

The alarming effects of the COVID-19 pandemic on different socio-economic spheres have been felt across the globe. These destructive effects have prompted plenty of research to understand and control the coronavirus pandemic. Notably, one strategic method of mitigating the effects of the coronavirus epidemic has been the utilisation of spatial and geostatistical models to gain insights into the potential predictors of the prevalence of the coronavirus. Considering the above, it was the aim of this study to explore the use of advanced geospatial modelling and analysis techniques, including Moran’s I, spatial error models, spatial lag models, MGWR, and GWR for analysing and modelling the settlement level determining factors of COVID-19 incidence within the eThekwini Metro to inform effectual micro-level planning. Notably, the lack of micro-level modelling of COVID-19 prevalence predictors also motivated the undertaking of this study. To the above aim, the objectives of the research were to utilise spatial autocorrelation to map the granular level COVID-19 spatial distribution over the 3rd wave in the eThekwini Metro, compare the applicability of global and local models in analysing and modelling micro-level COVID-19 incidence, analyse the spatial dependence of the occurrence of COVID-19 on local level variables through Moran’s I and to spatially model the effects of significant local-level determinants on COVID-19. The incidence of COVID-19 cases for the 3rd wave, which was from the 2nd of May 2021 to the 11th of September 2021, was analysed and modelled. The Moran’s I result illustrated that COVID-19 incidence within the eThekwini settlement places had a positive spatial autocorrelation, with a Moran’s I value of 0.14 and a p-value of 0.00. Also, the MGWR model's local R2 value was greater (72.5%) as compared to the other models. Moreover, economic wellness score, the sum of TB cases and population density came out as the significant determining factors of settlement level incidence of COVID-19. This research report offers a great foundation for gaining insights into the applicability of advanced geospatial models in guiding targeted COVID-19 interventions at lower levels.