Impact of the built environment and travel flow on COVID-19 Transmission: an Empirical Study of Georgia, United States

The built environment is often considered as a structural determinant of health while in such setting travel behavior also plays a particularly important role in the emergence and transmission of communicable disease. This study, using real data from multiple sources, is aimed to explore the association between COVID-19 transmission, the built environment, and travel flow at a localized scope. The conceptual framework of this study is based on an ecological model, which incorporates factors from the entire cycle while emphasizes urban-level and individual-level determinants. It begins with a time-dependent SIR model to quantitively measure and estimate state-level transmission rate by the method of ARIMAX in order to inform a big picture of the existing condition in terms of disease transmission. A foreseeable ‘turning point’ will only appear under the circumstance where total travel flow continues to drop. Then, it explores the spatial pattern of county-level COVID-19 transmission by methods including hierarchical clustering and Local Moran’s I with EB rate with respect to indicators such as cases per 100k, and average transmission rate. A various number of hotspots and coldspots are identified by different methods. Last but not least, a spatial regression model is set up to examine the association between disease transmission, the built environment, and travel flow, adjusted for socioeconomic status. Travel flow is quantified by M50 index, which represents the total amount of flow, and weighted network centrality, which measures the relative centrality of a geographical unit in predicted travel network combined with the risk of exposure to disease. The outcome shows that network centrality and density variance are positively associated with cases per 100k while population density has a negative impact, which is quite beyond common expectation. The contribution of this research is twofold. Firstly, a refined SIR model will help the local government to better predict the spike in order to get prepared in advance. This method will also be transferrable and applied to other states with corresponding data. Secondly, it tries to build up a framework to understand the association between disease transmission and selected characteristics of the physical and social environment, which will assist planning practitioners in future policy and spatial interventions to promote resiliency against health crisis like COVID-19.