Anonymization of Georeferenced Microdata
Rainer Lenz1, Simon Cremer1, Lydia Jehmlich* 1
Abstract
Spatial health data is becoming increasingly important in health research. However, the desired information can often not be extracted despite the inherent analytical content. The reason for this is that access to personal georeferenced data sets is severely restricted as they are subject to legal data protection. The method of donut masking attempts to alienate original data by shifting it in such a way that data protection is guaranteed without strongly reducing the analytical validity of the data. In this article donut masking is applied to partially synthetic data on sleep disorders. The degree of anonymity of the masked data set is measured by spatial k-anonymity reviewing additional knowledge of a potential data attacker. In addition to assessing the spatial similarity of the original and masked data set, an attempt is also made to assess the suitability of such data for analysis purposes.