生成式人工智能在药物研发中的应用、挑战与监管策略
收稿日期: 2025-06-25
修回日期: 2025-11-25
录用日期: 2026-02-06
网络出版日期: 2026-02-11
基金资助
药品监管科学国家重点实验室项目（2023SKLDRS0151）
Applications， Challenges and Regulatory Strategies of Generative Artificial Intelligence in Drug Development
Received date: 2025-06-25
Revised date: 2025-11-25
Accepted date: 2026-02-06
Online published: 2026-02-11
生成式人工智能（Gen AI）正在重塑药物研发的各个环节，包括靶点发现、分子设计、临床研究优化及注册文档生成。Gen AI通过深度学习模型分析多源数据，优化药物靶点识别，加速候选分子的设计与筛选，优化临床试验设计和患者招募。尽管Gen AI在提高研发效率和成功率方面展现出巨大潜力，但其应用仍面临可解释性不足、数据隐私保护与合规性、模型偏见、专业人才和跨界知识缺口、全球法规标准和指导原则滞后、成功案例难以复制等挑战。为应对这些问题，药品监管机构探索先行先试推动开展监管科学实践，坚持社会共治协同发力攻克技术难题，抓住发展机遇引领医药产业高质量发展，推动AI技术的法规支持和合规管理。通过政策引导与产业协同，Gen AI有望进一步推动药物研发的创新和生物医药产业的高质量发展。
侯健
,
陈小明
,
张原
,
李海玲
,
袁利佳
,
许真玉
.
生成式人工智能在药物研发中的应用、挑战与监管策略
Generative artificial intelligence （Gen AI） is reshaping various stages of the pharmaceutical development process， including target discovery， molecular design， clinical trial optimization， and regulatory document generation. By utilizing deep learning models to analyze multi-source data， Gen AI enhances drug target identification， accelerates the design and screening of candidate molecules， and optimizes clinical trial designs and patient recruitment. Although Gen AI shows tremendous potential in enhancing the efficiency and success rates of R&D， its application still faces challenges such as insufficient interpretability， data privacy protection and compliance issues， model bias， a shortage of specialized talent and interdisciplinary knowledge gaps， lagging global regulatory standards and guidelines， and difficulties in replicating successful cases.To address these issues， drug regulatory agencies are exploring pilot initiatives to advance regulatory science practices， promoting collaborative governance to tackle technical challenges， and seizing development opportunities to lead the high-quality growth of the pharmaceutical industry， thereby promoting regulatory support and compliance management for AI technologies. Through policy guidance and industrial collaboration， Gen AI is poised to further drive innovation in drug development and contribute to the high-quality advancement of the biopharmaceutical industry.
［1］ Zhang K， Yang X， Wang Y， et al. Artificial intelligence in drug development［J］.Nat Med， 2025，31（1）：45-59.
［2］ Jayatunga MKP， Xie W， Ruder L， et al. AI in small-molecule drug discovery： a coming wave？［J］.Nat Rev Drug Discov， 2022，21（3）：175-176.
［3］ Grand View Research. Artificial Intelligence In Drug Discovery Market Report， 2030［EB/OL］.（2024-07-26）［2025-05-04］.https：//www.grandviewresearch.com/industry-analysis/artificial-intelligence-drug-discovery-market.
［4］ 国家药品监督管理局.国家药监局综合司关于印发药品监管人工智能典型应用场景清单的通知［EB/OL］.（2024-06-18）［2025-11-02］.https：//www.nmpa.gov.cn/xxgk/fgwj/gzwj/gzwjzh/20240618144318144.html？type=pc&m=.
［5］ 工业和信息化部，商务部，国家卫生健康委，等.工业和信息化部等七部门关于印发《医药工业数智化转型实施方案（2025—2030年）》的通知［EB/OL］.（2025-04-24）［2025-11-02］.https：//www.nmpa.gov.cn/xxgk/fgwj/qita/20250425170354149.html.
［6］ Ren F， Aliper A， Chen J， et al. A small-molecule TNIK inhibitor targets fibrosis in preclinical and clinical models［J］.Nat Biotechnol， 2025，43（1）：63-75.
［7］ Richardson P， Griffin I， Tucker C， et al. Baricitinib as potential treatment for 2019-nCoV acute respiratory disease［J］.The Lancet， 2020，395（10223）：e30-e31.
［8］ FDA. Coronavirus （COVID-19） Update： FDA authorizes drug combination for treatment of COVID-19 ［EB/OL］.（2020-11-19）［2025-04-21］.https：//www.fda.gov/news-events/press-announcements/coronavirus-covid-19-update-fda-authorizes-drug-combination-treatment-covid-19.
［9］ Kamya P， Ozerov IV， Pun FW， et al. PandaOmics： an AI-driven platform for therapeutic target and biomarker discovery［J］.J Chem Inf Model， 2024，64（10）：3961-3969.
［10］ PandaOmics［EB/OL］.［2025-04-19］.https：//pharma.ai/pandaomics.
［11］ Ren F， Aliper A， Chen J， et al. A small-molecule TNIK inhibitor targets fibrosis in preclinical and clinical models［J］.Nat Biotechnol， 2025，43（1）：63-75.
［12］ Kingwell K. AI serves up target and inhibitor for lung fibrosis［J］.Nat Rev Drug Discov， 2024，23（5）：337-337.
［13］ Jusoh AS， Remli MA， Mohamad MS， et al. How generative artificial intelligence can transform drug discovery？［J］.Eur J Med Chem， 2025，295：117825.
［14］ Ragoza M， Masuda T， Koes DR. Learning a continuous representation of 3D molecular structures with deep generative models［EB/OL］.（2020-11-15）［2025-04-21］.https：arxiv.org/abs/2010.08687.
［15］ Zhou X， Cheng X， Yang Y， et al. DecompOpt： controllable and decomposed diffusion models for structure-based molecular optimization［EB/OL］.（2024-03-07）［2025-04-21］.https：//arxiv.org/abs/2403.13829.
［16］ Sousa T， Correia J， Pereira V， et al. Generative deep learning for targeted compound design［J］.J Chem Inf Model， 2021，61（11）：5343-5361.
［17］ Zeng X， Wang F， Luo Y， et al. Deep generative molecular design reshapes drug discovery［J］.Cell Rep Med， 2022，3（12）：100794.
［18］ Chemical. AI-AI for retrosynthesis & molecule synthesis［EB/OL］.［2025-05-31］.https：//www.chemical.ai.
［19］ Simulations Plus | Modeling & Simulation Software［EB/OL］.［2025-05-06］.https：//www.simulations-plus.com/.
［20］ ADMETlab 3.0 ［EB/OL］.［2025-05-06］.https：//admetlab3.scbdd.com/.
［21］ SwissADME ［EB/OL］.［2025-05-06］.http：//www.swissadme.ch/index.php.
［22］ Niazi SK， Mariam Z. Artificial intelligence in drug development： reshaping the therapeutic landscape［J］.Ther Adv Drug Saf， 2025， 16： 20420986251321704.
［23］ Zhang B， Zhang L， Chen Q， et al. Harnessing artificial intelligence to improve clinical trial design［J］.Commun Med， 2023，3（1）：1-3.
［24］ Dri DA， Massella M， Carafa M， et al. Artificial intelligence for drug product lifecycle applications［M］.2025： 205-234.
［25］ Coherent Solutions. Machine learning and AI in clinical trials： use cases ［2025］［EB/OL］.（2025-04-15）［2025-05-06］.https：//www.coherentsolutions.com/insights/role-of-ml-and-ai-in-clinical-trials-design-use-cases-benefits.
［26］ Pammi M， Shah PS， Yang LK， et al. Digital twins， synthetic patient data， and in-silico trials： can they empower paediatric clinical trials？［J］.Lancet Digit Health， 2025，7（5）：100851.
［27］ Bordukova M， Makarov N， Rodriguez-Esteban R， et al. Generative artificial intelligence empowers digital twins in drug discovery and clinical trials［J］.Expert Opin Drug Discov， 2024，19（1）：33-42.
［28］ Vidovszky AA， Fisher CK， Loukianov AD， et al. Increasing acceptance of AI‐generated digital twins through clinical trial applications［J］.Clin Transl Sci， 2024，17（7）：e13897.
［29］ Mann DL. The use of digital healthcare twins in early-phase clinical trials［J］.JACC Basic Transl Sci， 2024，9（9）：1159-1161.
［30］ Sanofi. Digital “twinning”： clinical trials powered by AI ［EB/OL］.（2024-05-22）［2025-05-06］.https：//www.sanofi.com/en/magazine/our-science/digital-twinning-clinical-trials-ai.
［31］ Wang Z， Gao J， Danek B， et al. InformGen： an AI copilot for accurate and compliant clinical research consent document generation［EB/OL］.（2025-04-01）［2025-04-21］.https：//arxiv.org/abs/2504.06934.
［32］ Markey N， El-Mansouri I， Rensonnet G， et al. From RAGs to riches： utilizing large language models to write documents for clinical trials［J］.Clin Trials Lond Engl， 2025：17407745251320806.
［33］ Lu X， Yang C， Liang L， et al. Artificial intelligence for optimizing recruitment and retention in clinical trials： a scoping review［J］.J Am Med Inform Assoc JAMIA， 2024，31（11）：2749-2759.
［34］ FDA.Research C for DE and. Considerations for the use of artificial intelligence to support regulatory decision-making for drug and biological products ［EB/OL］.（2025-01-06）［2025-05-06］.https：//www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-use-artificial-intelligence-support-regulatory-decision-making-drug-and-biological.
［35］ Intersoft consulting services AG. General Data Protection Regulation（GDPR）-Legal Text［EB/OL］.（2018-05-25）［2025-05-06］.https：//gdpr-info.eu/.
［36］ 中华人民共和国科学技术部.人类遗传资源管理条例实施细则［EB/OL］.（2023-06-01）［2025-05-03］.https：//www.most.gov.cn/xxgk/xinxifenlei/fdzdgknr/fgzc/bmgz/202306/t20230601_186416.html.
［37］ Pati S， Kumar S， Varma A， et al. Privacy preservation for federated learning in health care［J］.Patterns， 2024，5（7）：100974.
［38］ Hasanzadeh F， Josephson CB， Waters G， et al. Bias recognition and mitigation strategies in artificial intelligence healthcare applications［J］.Npj Digit Med， 2025，8（1）：1-13.
［39］ Salmi M， Atif D， Oliva D， et al. Handling imbalanced medical datasets： review of a decade of research［J］.Artif Intell Rev， 2024，57（10）：273.
［40］ Cross JL， Choma MA， Onofrey JA. Bias in medical AI： implications for clinical decision-making［J］.PLOS Digit Health， 2024，3（11）：e0000651.
［41］ Chawla NV， Bowyer KW， Hall LO， et al. SMOTE： synthetic minority over-sampling technique［J］.J Artif Intell Res， 2002，16：321-357.
［42］ Li C， Ding S， Zou N， et al. Multi-task learning with dynamic re-weighting to achieve fairness in healthcare predictive modeling［J］.J Biomed Inform， 2023，143：104399.
［43］ Catherina X， Tulsee Doshi. Fairness indicators： scalable infrastructure for fair ML systems［EB/OL］.（2019-12-11）［2025-05-17］.https：//research.google/blog/fairness-indicators-scalable-infrastructure-for-fair-ml-systems/.
［44］ Raftopoulos G， Fazakis N， Davrazos G， et al. A Comprehensive review and benchmarking of fairness-aware variants of machine learning models［J］.Algorithms， 2025，18（7）：435.
［45］ Guo LL， Pfohl SR， Fries J， et al. Systematic review of approaches to preserve machine learning performance in the presence of temporal dataset shift in clinical medicine［J］.Appl Clin Inform， 2021，12（4）：808-815.
/
| 〈 |
|
〉 |
