摘要
互联网技术在医疗领域的广泛应用,带动了数字医疗产业的飞速发展,也促使大量医学人工智能产品出现。由于人工智能产品部署的特殊性,产品的网络安全的问题越来越受到关注。这些问题包括如何在产品设计、研发、数据集构建、算法训练等过程中考虑网络安全因素,保障性能的可靠性、模型健壮性等要素,如何在产品全生命周期过程中从通用安全和专用安全的角度采取相关策略防御网络安全风险。本文针对通用安全和专用安全开展研究,具体分析了通用安全的四个问题和专用安全的三个问题及防御机制,为保证医学人工智能产品全生命周期的安全性,提供建议和思考。
With the world-wide application of Internet technology in the medical field,the development of digital medical industry has been promoted rapidly.Especially,this technique urges the emergence of a large number of medical artificial intelligence products.Due to the particularity of the deployment mode for such products,the cyber security of products has drew more and more attentions.These problems include how to consider network security factors in the process of product design,research and development,data set construction,algorithm training,etc.,how to guarantee the reliability of performance,model robustness and other elements,and how to adopt relevant strategies to defend network security risks from the perspective of general safety and special safety in the whole life cycle of products.In this study,we analyzed the four questions existed in generic security and three questions in special safety,as well as the defence mechanism.Our results will ensure the safety of medical AI products in life cycle.
作者
王晨希
王浩
王权
任海萍
WANG Chenxi;WANG Hao;WANG Quan;REN Haiping(Division of Active Medical Device and Medical Optics, National Institutes for Food and Drug Control, Beijing 102629, China)
出处
《中国医疗设备》
2018年第12期22-25,30,共5页
China Medical Devices
基金
国家重点研发计划项目(2016YFC0107100)
体育总局重点课题联合中国红十字基金会燎原基金项目(2015B101)
关键词
人工智能
网络安全
全生命周期
防御机制
artificial intelligence (AI)
cyber security
whole life cycle
defence mechanism
