ABSTRACT
Infectious diseases such as COVID-19 have severe impacts on both economy and public health in the US and the world. Due to the heterogeneity of virus spread, there are spatial variations in the demand for medical resources such as personal protective equipment (PPE), testing kits, and vaccines. The availability of such medical resources is critical to effective epidemic control. Although these resources can be readily transported to designated areas for fighting an epidemic, the demand is increasing and varying in space that places significant stress on the supply and allocation of medical resources. However, little has been done on the tessellation of infection distributions for resource management. In this letter, we develop new tessellation algorithms for decision support in epidemic resource allocation and management. The objective is to estimate resource locations and coverage based on the spatial analysis of heterogeneous infection distribution. First, spatial tessellation centroids are initialized through either greedy or cluster-centric approaches. Next, the locations of tessellation centroids are calibrated through a gradient learning algorithm. Lastly, the spread tessellation is computed to provide an estimation of resource coverages under the heterogeneous infection distribution. The proposed methodology is evaluated and validated using a COVID-19 case study of infection data in Pennsylvania. Experimental results show the proposed methodology effectively tessellates the spread of infectious diseases. The new spread tessellation algorithms are shown to have strong potentials for epidemic decision support in infection modelling and resource allocation.