摘要
ARX工具是一款由德国慕尼黑工业大学开发,具有高效的全域匿名化算法、丰富的隐私模型、数据效用评估方法和重识别风险分析功能,被许多研究者和隐私政策制定者推荐的开源生物医学数据匿名化工具。通过梳理ARX工具的发展历史和相关研究,将ARX工具的功能架构归纳为数据导入导出模块、数据处理模块、隐私模型模块及其他参数模块、匿名化方案探索模块、效用分析模块及风险分析模块5个部分,并以Adult数据集的匿名化为例梳理了ARX工具的工作流程,总结形成5个步骤。然后通过与其他匿名化工具的功能进行比较和分析,归纳了ARX工具的4个特点并总结了各模型和指标的原理和含义及ARX工具应用的局限性。最后结合我国数据共享的实际需求,针对性地提出了我国数据匿名化工具研发时需要重点关注的内容,对我国生物医学数据匿名化工具的研发具有一定的参考价值。
ARX is an open source biomedical data anonymization tool,developed by Technical University of Munich,with efficient full-domain anonymization algorithms,a wide variety of privacy models,data utility assessment methods and re-identification risk analysis function and recommended by many researchers and privacy policy makers.By reviewing the history of ARX and sorting out the related studies,the functional architecture of ARX was summarized into five parts:data import and export module,data processing module,privacy model and other parameters module,anonymization scheme exploration module,utility analysis and risk analysis module.The workflow of the ARX consisted of five steps using the anonymization of Adult datasets as an example.By analyzing and comparing the functions of ARX with those of other anonymization tools,four characteristics were summarized,and the principles and meanings of various models and indicators as well as the limitations of ARX were also summarized.Combined with the actual needs of data sharing in China,the content to be focused on was put forward to provide reference for the research and development of biomedical data anonymization tools in China.
作者
唐明坤
钱庆
张丽鑫
周佳茵
吴思竹
TANG Ming-kun;QIAN Qing;ZHANG Li-xin;ZHOU Jia-yin;WU Si-zhu(Institute of Medical Information,Chinese Academy of Medical Sciences&Peking Union Medical College,Beijing 100020,China)
出处
《中华医学图书情报杂志》
CAS
2022年第2期19-29,共11页
Chinese Journal of Medical Library and Information Science
基金
中国医学科学院项目“医学与健康科技创新工程”(2021-I2M-1-057)
中国医学科学院项目“国家人口健康科学数据中心后补助课题”(NCMI-KD01N-202109)。
关键词
数据共享
数据匿名化
统计披露控制
ARX
Data sharing
Data anonymization
Statistical disclosure control
ARX
