Using an Eigenvector Spatial Filtering Based Spatially Varying Coefficient Model to Analyze the Spatial Heterogeneity of COVID-19 and Its Influencing Factors in Mainland China
Abstract Background: The COVID-19 pandemic has led to many deaths and economic disruptions across the world. Several studies have examined the effect of health risk factors on COVID-19 rates in different places, but the problem of spatial heterogeneity has not been adequately addressed.Methods: In this paper, we developed an Eigenvector Spatial Filtering based spatially varying coefficient model (ESF-SVC) to reveal the spatially varying impact of certain health risk factors on the COVID-19 spread. The experiment was conducted during 7 weeks within two study extents (Hubei province and mainland China). Spatial varying coefficient maps were produced for spatial pattern discovery.Results: Results showed that the ESF-SVC model could take good control of over-fitting problems, with average adjusted R2 16.31% (in Hubei province) and 10.25% (in mainland China) higher than that of GWR. The cross validation RMSE of ESF-SVC model was also the lowest. In Hubei province, Population density and wind speed had a significant impact on COVID-19 infection rates and that their effect was constant across cities. While in mainland China, migration score, building density, temperature and altitude showed significant impact and their effect varies across space. The influence of migration score was less contributive and less significant in cities around Wuhan than cities farther away, while the altitude showed stronger contributions in high altitude cities.Conclusions: Our study hopes to provide not only a feasible path to solve the problem of spatial autocorrelation and spatial heterogeneity in COVID-19 characterization but also an intuitive way to discover spatial patterns in large study areas, which could help people and government awareness of the potential health risks and shed some light in COVID-19 control strategies.