Geo-Clustering Model for Optimizing Locations of Public Health Emergency Operations and COVID-19 Vaccine Distribution Centers

Optimum location of vaccine distribution and Emergency Operation Centers (EOCs) is imperative to ensuring prompt and efficient vaccination of eligible population in any location of interest. The proximity of these vaccination centers is likely to positively affect the decision of the target population to present themselves for vaccination. In this paper, a computational model for optimizing the number and determining the location of depots or vaccine distribution centers, and amounts of vaccines to be stocked at each center, to satisfy the needs of the local population is proposed. A modified K-means++ is used to optimize the number of required centers and the approximate locations to ensure the usage of the least possible cost. The algorithm allows planners to enter two initial specific locations as depots, thereby avoiding the usual random selection of initial points. Using geospatial and population data, the resulting clusters are divided into two, on each iteration. Heap sort is used to select the next centroid. Optimization of these locations is iteratively done, until there are no more changes. An optimized number of vaccine distribution centers for any region of interest can be obtained. It ensures that least possible cost is used. Our algorithm avoids the usual random outcomes associated with K-means and provides a more efficient clustering output, with an improved time complexity. The application of the proposed algorithm to a real-world test instance indicates its effectiveness.