Microscale mapping of COVID-19 cases: A kernel-based clustering approach

Keywords: COVID-19, RCMC, aggregate, spatial uncertainty
Session Type: Virtual Paper Abstract
Day: Saturday
Session Start / End Time: 2/26/2022 09:40 AM (Eastern Time (US & Canada)) - 2/26/2022 11:00 AM (Eastern Time (US & Canada))
Room: Virtual 4

Abstract

The onset and progression of the COVID-19 pandemic saw the creation of several maps visualizing cases and case rates aggregated to various geographies such as ZIP codes, towns and counties. While certainly useful in painting a big picture of public health outcomes at specific aggregate scales, such mapping efforts may inadvertently hide nuanced patterns and clusters of the disease on a microscale community level due to the inherent assumption that cases are distributed evenly across the aggregate area. As such, there may be a significant delay in efforts towards policies that ensure access to quality health services for vulnerable populations in such communities.
This paper will discuss a kernel-based disease mapping method that disaggregates town-level location data into probabilistic point-level locations using the Restricted and Controlled Monte Carlo (RCMC) process. Generally, the kernel-based approach explicates the heterogeneous background over which to estimate the intensity of disease, by calculating the ratio between COVID-19 cases and the population at risk. The RCMC assigns town-level cases to random point locations. This randomization is restricted by the original town boundaries and is controlled by a high-resolution population dataset so that the probability of a location to receive a case of the randomization is proportional to the population value at that location. The variance in the scenarios based on several simulations represents the spatial uncertainty sourced from the town-level location information. Finally, microscale gridded cluster maps are visualized to highlight communities that could benefit from urgent health policy priorities such as vaccination program targeting.