PDF下载

[1]鞠鑫姚永刚赵芯蕊.医疗健康领域人工智能应用安全框架研究[J].中国卫生质量管理,2026,33(5):7-11.[doi:10.13912/j.cnki.chqm.2026.33.5.02]
　JU Xin,YAO Yonggang,ZHAO Xinrui.Research on the Security Framework for Artificial Intelligence Applications in the Medical and Health Field[J].Chinese Health Quality Management,2026,33(5):7-11.[doi:10.13912/j.cnki.chqm.2026.33.5.02]

点击复制

医疗健康领域人工智能应用安全框架研究

《中国卫生质量管理》[ISSN:1006-7515/CN:CN 61-1283/R] 卷: 33 期数: 2026年5期页码: 7-11 栏目: 特别关注出版日期: 2026-05-15

Title:: Research on the Security Framework for Artificial Intelligence Applications in the Medical and Health Field

作者:: 鞠鑫姚永刚赵芯蕊; 苏州市卫生健康信息中心

Author(s):: JU Xin; YAO Yonggang; ZHAO Xinrui; Suzhou Health Information Center

关键词:: 医疗领域; 人工智能大模型; 可信执行环境; 风险检测大模型; 安全框架

Keywords:: Medical Field; Artificial Intelligence Large Models; Trusted Execution Environment; Risk Detection Large Models; Security Framework

分类号:: R197.3

DOI:: 10.13912/j.cnki.chqm.2026.33.5.02

文献标志码:: B

摘要:: 随着AI技术的快速发展，大模型在医疗健康领域的应用潜力日益凸显，医疗数据的高敏感性和隐私保护需求也对大模型全生命周期安全管控提出了更高要求。本研究围绕医疗健康领域AI大模型在私有云环境中的应用场景，提出了一套从大模型底座、运行环境到具体应用的全生命周期安全管控框架，主要包含数据采集、大模型训练、大模型推理和应用环节输入输出4个阶段的安全架构设计。通过引入多项国内外前沿大模型安全技术，包括数据隔离保护机制、模型攻击防护算法、隐私计算融合方案以及运行时动态风险监控等，全面提升了大模型安全性能，实现了医疗健康AI大模型安全风险闭环管理，为医疗健康领域AI应用安全管理提供了参考。

Abstract:: With the rapid development of artificial intelligence (AI) technology, the application potential of large models in the healthcare field has become increasingly prominent. The high sensitivity of medical data and the demand for privacy protection have also imposed higher requirements on the full lifecycle security management and control of large models. Focusing on the application scenarios of AI large models in the healthcare field within private cloud environments, this study proposes a full lifecycle security management and control framework spanning from the foundation of large models and their operating environments to specific applications. It primarily encompasses security architecture design across four stages: data collection, large model training, large model inference, and input or output in application scenarios. By introducing multiple internationally and domestically leading security technologies for large models, including data isolation protection mechanisms, model attack prevention algorithms, privacy computing integration solutions, and runtime dynamic risk monitoring, the overall security performance of large models has been comprehensively enhanced. This achieves closed-loop management of security risks for AI large models in healthcare, providing a reference for AI application security management in the healthcare field.

参考文献/References:

［1］付志远,陈思宇,陈骏帆,等.大语言模型安全的挑战与机遇［J］.信息安全学报,2024(5):26-55.［2］肖非易,李雪,李睿,等.医疗人工智能技术评估与监管的国际经验及启示［J］.中国卫生质量管理.2023,30 (7):58-62.［3］杨晓姣,罗仙,张玲.从美智库报告看人工智能大语言模型网络安全问题及对策［J］.信息安全与通信保密,2024 (8):20-29.［4］PRANAV R,EMMA C,OISHI B,et al. AI in health and medicine［J］.Nat Med,2022,28(1),31-38.［5］林梓瀚,郭丰.人工智能时代我国数据安全立法现状与影响研究［J］. 互联网天地,2020(9):20-25.［6］周吉银,刘丹,曾圣雅.人工智能在医疗领域中应用的挑战与对策［J］.中国医学伦理学,2019,32 (3):281-286.［7］王乔晨,吴振刚,刘虎.大语言模型应用的安全与隐私问题综述［J］.工业信息安全,2024(5):40-45.［8］孙磊,汪安安,宋一敏,等.大语言模型在临床医学领域的应用、挑战和展望［J］.解放军医学院学报,2025,46(1):50-60.［9］赵月,何锦雯,朱申辰,等.大语言模型安全现状与挑战［J］. 计算机科学,2024,51(1):68-77.［10］皮勇,张明诚.总体国家安全观视域下人工智能安全风险治理研究［J］.中国科技论坛,2023(6):86-96.［11］WANG C, CAO N, REN K,et al.Enabling secure and efficient ranked keyword search over outsourced cloud dat［J］. IEEE Transactions on Parallel and Distributed Systems,2012,23(8):1467-1479.［12］KAIROUZ EBP, MCMAHAN HB, AVENT B,et al. Advances and open problems in federated learning［J］.Foundations and Trends in Machine Learning,2021,14(1):1-2.［13］池亚平，吴冰，徐子涵.大语言模型在威胁情报生成方面的研究进展［J］.信息安全研究,2024,10(11):1028-1035.［14］潘伟华,许健,穆嘉盛,等.应用人工智能提升单病种质量管理水平［J］.中国卫生质量管理,2022,29 (7):19-21.［15］胡振生,杨瑞,朱嘉豪,等.大语言模型在医学领域的研究与应用发展［J］.人工智能,2023(4):10-19.

更新日期/Last Update: 2026-05-15