Protecting patient geo-privacy via a triangular displacement geo-masking method

Abstract

Protecting patient geo-privacy while allowing for valid geographic analyses of the data is a major challenge [1]. As a consequence, a variety of methods have been introduced to mask patients' locational information, also called geo-masking methods [2]. This study assessed the five main geo-masking methods as cited by [3] in terms of re-identification risk and performance. These five methods are Random Direction and Fixed Radius, Random Perturbation within a Circle, Gaussian Displacement, Donut Masking, and Bimodal Gaussian Displacement. Based on the assessment, the study highlighted two major gaps in the design of these geo-masking methods. First, all five geo-masking methods used only population density in calculating the displacement distances between the original locations of points and their new locations. However, other criteria that might be as important as population density were not considered in designing these five methods. These include data sensitivity, research types, quasi-indicator availability, previously generated maps availability, end-users' types, and the possibility of temporal synergy of data. Second, the Donut Masking and the Bimodal Gaussian Displacement methods were found to be superior in terms of minimizing re-identifying risks, but they were also found to be consuming much more processing power compared to the other three geo-masking methods. To address these gaps, this study proposed a new geo-masking method, called the "Triangular Displacement". The primary design, development, and evaluation of the Triangular Displacement method were based on the Design Science Research (DSR) Process Model [4], also known as DSRM. The expected next step is to implement the resultant geomasking method as a tool to help healthcare data guardians deidentify large sets of PHR automatically. A pilot study with a large Southern Californian healthcare provider has been outlined to examine the efficacy of the developed solution.

An Evaluation Framework for Assessing the Impact of Location Privacy on Geospatial Analysis

“…Exchanges the mobility traces of two different individuals for a certain period of time. Geographic coordinates these are presented in [15], where density is had under consideration in the methods. in [2] is described displacement, change of scale, rotation.…”

Methods for deriving and calibrating privacy-preserving heat maps from mobile sports tracking application data

“…Still, an open question remains whether the value k should be a function of the density of the population or the underlying road network to protect the confidentiality of sensitive locations even in areas where there are small variations in the routes. A similar issue has been discussed in many geographical masking studies that have suggested weighting the displacement distance by population density because in less densely populated areas the risk of re-identification is higher (e.g., Armstrong and Ruggles, 2005;Kwan et al, 2004;Murad et al, 2014;Seidl et al, 2015).…”

Decentralized geoprivacy: leveraging social trust on the distributed web

“…Geo-masking (Seidl et al 2016, Zhang et al 2017 Adaptive geo-masking, Voronoibased aggregation system (Croft et al 2016) Triangular displacement (Murad et al 2014) k-anonymity (Samarati andSweeney 1998, Sweeney 2002). Point aggrigation and adaptive arial elimination.…”