Quantifying fine-grained privacy risk and representativeness in medical data
Xiaoqian Jiang, Samuel S. M. Cheng, Lucila Ohno‐Machado · 2011
The current way that privacy is protected suggests lack of consideration of privacy needs of different types of medical records. Without knowing how representative and sensitive individual records are, methods built on the top of unwarranted assumptions could be misleading and even erroneous, e.g. over-protecting of a subset of the records while under-protecting of the rest. This article developed a novel framework to quantify the fine-grained privacy risk and representativeness of individual records in a medical database. Our implementation leveraged the KD-tree, an efficient data structure, to do range queries. We used real-data to demonstrate the feasibility of the proposed method.