A Quantitative Analysis of the Performance and Scalability of De-identification Tools for Medical Data

摘要: Recent developments in data de-identification technologies offer sophisticated solutions to protect medical when, especially the is be provided for secondary purposes such as clinical or biomedical research. So determine what degree an approach--- along with its tool--- usable and effective, this paper takes into consideration a number of tools that aim at reducing re-identification risk published data, yet preserving statistical meanings. We therefore evaluate residual by conducting experimental evaluation most stable research-based tools, applied our Electronic Health Records EHRs database, assess which tool exhibits better performance different quasi-identifiers. Our criteria are quantitative opposed other descriptive qualitative assessments. notice on comparing individual disclosure information loss each μ-Argus performs better. Also, generalization method considerably than suppression terms avoiding loss. also find sdcMicro has best scalability among counterparts, been observed experimentally virtual consisted 33 variables 10,000 records.