文献速递免疫规划卫生政策与健康促进传染病防控与疫情报告

数字孪生与可解释AI贯穿疫苗全生命周期:从mRNA生产到热稳定性、冷链物流与免疫服务

Digital twins and explainable AI across the vaccine life cycle: from mRNA manufacturing to thermostability, cold-chain logistics, and immunisation services.
PubMed 文献 · Vaccine · 2026-09-07

该叙述性综述检索2019-2025年多个数据库,综合50余项研究,分析数字孪生(DT)与可解释人工智能(XAI)在疫苗全生命周期中的应用。证据显示DT正从静态模拟转向质粒到mRNA生产的操作型孪生,XAI改善稳定性预测,AIoT强化冷链监控与路线规划;但当前实施常孤立、验证不足且少考虑公平性。

要点

  • DT正从静态模拟转向质粒到mRNA生产的操作型孪生,支持虚拟调试与实时优化。
  • XAI改善稳定性预测,AIoT架构强化冷链监控与路线规划。
  • 大规模接种中心与初级免疫服务的孪生可改善吞吐量与温度管理。

核心结论

  • DT与XAI在疫苗全生命周期中展现潜力,但当前实施常孤立、验证不足且少考虑公平性。
  • 领域优先事项包括整合多尺度的「疫苗系统孪生」、严格验证、有效人机协作及针对中低收入国家的部署策略。

研究设计与样本

  • 叙述性综述,检索PubMed、Web of Science、Scopus、IEEE Xplore和ACM Digital Library(2019-2025年),综合50余项研究。

主要发现

  • DT正从静态模拟转向质粒到mRNA生产的操作型孪生,支持虚拟调试、快速扩产和实时优化。
  • XAI方法改善稳定性预测,AIoT架构强化冷链监控与路线规划。
  • 大规模接种中心和初级免疫服务的孪生可改善吞吐量和温度管理。
  • 当前实施常孤立、验证不足且很少考虑公平性。

临床怎么用

  • 可利用DT支持mRNA生产虚拟调试与实时优化,借助XAI预测稳定性,通过AIoT加强冷链监控与路线规划。
  • 大规模接种中心与初级免疫服务可应用孪生改善吞吐量与温度管理,但需注意验证与公平性。

局限性

  • 当前DT与XAI实施常孤立、验证不足且很少考虑公平性。
  • 证据多来自叙述性综述综合,缺乏针对中低收入国家情境的部署策略验证。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

The COVID-19 era accelerated the deployment of mRNA vaccines while exposing persistent weaknesses in manufacturing, thermostability and cold-chain systems, particularly in low- and middle-income countries (LMICs). This narrative review, informed by an explicit search of PubMed, Web of Science, Scopus, IEEE Xplore and the ACM Digital Library (2019-2025), synthesises over 50 studies of digital twins (DTs) and explainable artificial intelligence (XAI) across the vaccine life cycle. The evidence indicates that DTs are progressing from static simulations to operational twins for plasmid-to-mRNA production, supporting virtual commissioning, rapid scale-up and real-time optimisation. In parallel, XAI-enabled approaches improve stability prediction, and artificial intelligence of things (AIoT) architectures strengthen cold-chain monitoring and route planning. At the service level, twins of mass vaccination centres and primary-care immunisation services can improve throughput and temperature management; however, current implementations are often siloed, weakly validated and rarely designed with equity considerations. Priorities for the field include integrated, multi-scale 'vaccine system twins', rigorous validation, effective human-AI collaboration and deployment strategies tailored to LMIC contexts.

来源:Vaccine,查看原文。版权归原作者所有。

引用

  1. Ugwu OP, Okon MB, Ogenyi FC, et al. Digital twins and explainable AI across the vaccine life cycle: from mRNA manufacturing to thermostability, cold-chain logistics, and immunisation services. Vaccine. 2026. PMID: 42705167. doi:10.1016/j.vaccine.2026.129131

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。