基于风险原生的药品智慧监管新架构研究
收稿日期: 2026-05-18
修回日期: 2026-06-11
录用日期: 2026-06-26
网络出版日期: 2026-08-17
Research on New Architecture for Intelligent Pharmaceutical Regulation Based on Risk-Native Design
Received date: 2026-05-18
Revised date: 2026-06-11
Accepted date: 2026-06-26
Online published: 2026-08-17
目的：在传统药品智慧监管架构面临“事后响应”“工具滞后”等困境下，破解基于风险的监管理念难以有效落地的难题，弥合理念先进与技术工具落后之间的架构鸿沟，探索监管现代化的新路径。方法：通过构建“监管仿真沙盒”验证框架，设计“数据基础层-风险智能层-监管应用层”的技术架构，对AI模型治理、可解释性及持续监控等关键技术问题进行系统阐述。结果：提出风险原生设计理论范式，在架构层面系统阐述风险识别、评估、预警与处置的内生实现机制。基于对美国FDA知识辅助评估与结构化分析（FDA KASA）项目等单环节风险原生实践的层次分析，明确从环节级风险原生向系统级风险驱动监管跃迁的实现路径。结论：风险原生设计范式为药品智慧监管提供了兼具理论创新性与工程可行性的解决方案，为我国药品监管现代化转型提供了有力的理论支撑与清晰的技术路径。
锋
,
由玉伟
.
基于风险原生的药品智慧监管新架构研究
Objective： To solve the difficulty of effectively implementing risk-based regulation concept under the dilemmas of "post-event response" and "tool lag" faced by traditional intelligent pharmaceutical regulatory architectures， to bridge the architectural divide between advanced concepts and outdated technological tools， and to explore a new pathway for regulatory modernization.Methods： A "regulatory simulation sandbox" validation framework was constructed， a three-tier technical architecture comprising the "data foundation layer， risk intelligence layer， and regulatory application layer" was designed， and key technical issues including AI model governance， interpretability， and continuous monitoring were systematically addressed.Results： The study proposes the “risk-native-design” theoretical paradigm and architecturally elucidates the endogenous implementation mechanisms for risk identification， assessment， early warning， and disposition. Based on a hierarchical analysis of single-link risk-native-design practices， such as the FDA's KASA （knowledge-aided assessment and structured application） initiative， a clear transition pathway from "link-level" risk embedding to "system-level" risk-driven regulation was formulated.Conclusion： The "risk-native-design" paradigm offers a solution that combines theoretical innovation with engineering feasibility for intelligent pharmaceutical regulation. It provides theoretical support and a clear technical pathway for the modernization transformation of pharmaceutical regulation in China.
［1］ 全国人民代表大会常务委员会.中华人民共和国药品管理法［EB/OL］. （2019-08-26）［2026-06-15］.http：//www.npc.gov.cn/npc/c2/c30834/201908/t20190826_300489.html.
［2］ International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. ICH Q9（R1） Quality Risk Management［EB/OL］.（2023-01-23）［2026-06-15］.https：//database.ich.org/sites/default/files/ICH_Q9%28R1%29_Guideline_Step4_2025_0115_0.pdf.
［3］ Food and Drug Administration. Advancing Regulatory Science at FDA： Focus Areas of Regulatory Science （FARS）［EB/OL］.（2021-10-01）［2026-06-15］.https：//www.fda.gov/science-research/science-and-research-special-topics/advancing-regulatory-science.
［4］ Scholl B， Swanson T， Jausovec P. Cloud native： using containers， functions， and data to build next-generation applications［M］.Sebastopol， CA： O'Reilly Media， 2022.
［5］ Food and Drug Administration.Knowledge-aided Assessment & Structured Application （KASA）：A New Approach that Modernizes FDA’s Quality Assessment of Regulatory Drug Applications［EB/OL］.（2022-11-03）［2026-06-15］.https：//www.fda.gov/media/162865/download？attachment.
［6］ Yang Q， Liu Y， Chen TJ， et al. Federated machine learning： concept and applications［J］.ACM Trans Intell Syst Technol， 2019，10（2）：1-12，19.
［7］ Dwork C， Roth A. The algorithmic foundations of differential privacy［M］.Boston， MA： Now Publishers， 2014.
［8］ Lundberg SM， Lee SI. A unified approach to interpreting model predictions［C］//Proceedings of the 31st International Conference on Neural Information Processing Systems （NeurIPS 2017）.Long Beach， CA： Curran Associates， 2017，30：4765-4774.
［9］ Ribeiro MT， Singh S， Guestrin C. 'Why should I trust you？'： Explaining the predictions of any classifier［C］//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining （KDD ’16）. San Francisco， CA： ACM， 2016：1135-1144.
［10］ European Medicines Agency. Data Analysis and Real World Interrogation Network （DARWIN EU）［EB/OL］.（2022-02-09）［2026-06-10］.https：//www.ema.europa.eu/en/about-us/how-we-work/big-data/data-analysis-real-world-interrogation-network-darwin-eu.
［11］ Food and Drug Administration. Sentinel Initiative： A national system for monitoring the safety of FDA-regulated medical products［EB/OL］.（2008-05-01）［2026-06-10］.https：//www.fda.gov/safety/fdas-sentinel-initiative.
［12］ 宋姗姗，黄璜，韦林沙.数字政府的系统框架构建与成熟度评估研究——基于“结构-过程-功能”的分析框架［J］.图书馆学研究，2025，（10）：40-50.
［13］ European Medicines Agency. Guiding principles on the use of large language models in regulatory science and for medicines regulatory activities［EB/OL］.（2024-08-29）［2026-06-10］.https：//www.ema.europa.eu/en/documents/scientific-guideline/guiding-principles-use-large-language-models-regulatory-science-medicines-regulatory_en.pdf.
［14］ Food and Drug Administration. Considerations for the use of artificial intelligence to support regulatory decision making for drug and biological products［EB/OL］.（2025-01-01）［2026-06-10］.https：//www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological.
/
| 〈 |
|
〉 |
