Privacy Preservation of Cluster Integrity on Web-Scraped Hospital Data

This paper proposes a method for anonymizing geospatial healthcare data that preserves the natural clustering of locations. A dataset is built with each healthcare facility's location, ratings, and review count by scraping it off an online maps platform. This enables the calculation of each facility's regional influence. The approach retains spatial clusters by restricting the displacements on each facility's geospatial coordinates to at most the minimum radius of the cluster while also factoring for the prominence of the facility in the region. This ensures that the shape and magnitude distribution of clusters is maintained, thereby preserving dataset quality. Impactful insights can be drawn from this anonymized dataset without having to identify any individual facility.