A Web-based Geographic Framework to Detect and Visualize Space-time Clusters of Infectious Diseases

Infectious diseases pose a significant threat to public health worldwide as evidenced by the recent coronavirus 2019 (COVID-19) pandemic. Despite significant human losses, the advent of web-accessed, map-based “data dashboards” that can monitor disease outbreaks, proved essential in managing public health responses. In many cases, the backend of these dashboards employs basic mapping functionality, displaying counts or rates. As the pandemic advanced, the identification of elevated rates was increasingly important in the geographical allocation of public health resources. However, such maps miss the opportunity to provide accurate information to policy decision makers such as the rate of disease spread, cyclicity, direction, intensity, and the risk of diffusion to new regions. Space-time geoanalytics, when coupled with rich visualizations, can address these shortcomings. Moreover, when implemented over the web, such functionality can be accessed from virtually anywhere.
This dissertation presents a web-based geographic framework for detecting and visualizing explicit space-time clusters of infectious diseases. First, I conduct a systematic review of the literature around the theme of space-time cluster detection for infectious diseases to identify state-of-the-art techniques that should be included in the proposed web-based framework. Second, I develop a tightly coupled, web-based analytical framework for the detection of clusters of infectious diseases using interactive and animated 3D visualizations to aid epidemiologists in readily and adequately uncovering the characteristics of space-time clusters. As a proof of concept, I populate the framework with COVID-19 county-level data for the 48 contiguous states in the US, and demonstrate data retrieval and storage, space-time cluster detection analysis, and 3D visualization within an open source WebGIS environment. Third, I evaluate the prototype in two steps: 1) present this and two existed COVID-19 systems to a group of infectious diseases experts and solicit feedback, 2) and evaluate functionalities on the prototype by conducting a user study with graduate students in a setting of online surveys.
This tightly coupled approach facilitates the detection of space-time clusters of diseases in a computationally acceptable timeframe. The characteristics of this framework (generic, open source, highly accurate, modifiable) will enable low-cost monitoring of the spatial and temporal trends of diseases causing high risks of infection.