为医疗人工智能赢得信任：协作前行之路

Earning Trustfor AI in Health:A CollaborativePath Forward W H I T EP A P E RJ U N E2 0 2 5 Contents Foreword3Executive summary4Introduction51Empowering trustworthy AI in health: The urgent need for collaboration71.1Global divergences challenge the scaling of AI in health81.2The private sector is key to driving progress and standardization81.3AI regulations must be crafted to keep pace with innovation92The need for a pragmatic approach: Guidelines, sandboxes and10post-market surveillance2.1Legislation can build a strong baseline for governing AI in health102.2Sandboxes provide a safe space in which the private11sector can innovate2.3Post-market surveillance can help cope with12the evolving nature of AI3The importance of public–private partnerships for AI in health133.1The role of public–private partnerships in regulating13medical devices, including software3.2Private-sector capabilities can help test and13operationalizethe regulatory process3.3Quality assurance resources: An approach to PPPs15for independent testing and trainingConclusion16Appendix: A selection of regulatory sandbox inititatives17Contributors18Acknowledgements18Endnotes19 Disclaimer This document is published by theWorld Economic Forum as a contributionto a project, insight area or interaction.The findings, interpretations andconclusions expressed herein are a resultof a collaborative process facilitated andendorsed by the World Economic Forumbut whose results do not necessarilyrepresent the views of the World EconomicForum, nor the entirety of its Members,Partners or other stakeholders.©2025 World Economic Forum. All rightsreserved. No part of this publication maybe reproduced or transmitted in any formor by any means, including photocopyingand recording, or by any informationstorage and retrieval system. Foreword Andy Moose Head of Health andWellness, Centre forHealth and Healthcare,World Economic Forum Ben HornerManaging Directorand Partner,Boston Consulting Group Artificial intelligence (AI) holds great promise totransform healthcare – enhancing diagnostics,optimizing workflows and improving healthoutcomes for all. However, realizing AI’s benefitsresponsibly demands a fundamental evolutionof how health systems, with their diverse set ofstakeholders, develop and build trust in innovation. Equally important is strengthening technicalcapacity among regulators, innovators andhealthcare leaders to develop a sharedunderstanding of AI’s capabilities and risks.Public–private partnerships should be positionedat the core of this transformation – co-developingstandards, supporting regulatory innovation andbuilding shared infrastructures for evaluation andmonitoring. Strong international collaborationwill be critical to harmonize approaches,foster interoperability and enable scaling of AItechnologies across health systems. Existing evaluation frameworks – built for productsthat remain typically unchanged after approval,such as pharmaceuticals and medical devices –are not fully equipped to manage the dynamic,evolving nature of AI technologies. The probabilisticbehaviour of certain AI systems introducesnew dimensions of uncertainty that traditional,deterministic approaches cannot fully address. If we act now, we can embed trust in thefoundations of digital health transformation. Byaligning innovation with ethical principles andfocusing on continuous evaluation, AI can fulfil itspromise: improving health outcomes, enhancingsystem resilience and expanding access to high-quality care throughout populations. To manage these challenges effectively, regulatorymodels must evolve. Dynamic governancemechanisms such as regulatory sandboxes, life-cycle evaluation and post-market monitoring willbe essential to ensure that AI systems remain safe,effective and equitable throughout their lifespan.Complementary to legislation, guidelines can helpmaintain innovation while setting clear societalguardrails and industry standards. Executive summary AI will reshape healthcare, but realizingits full potential requires responsiblegovernance, trust and global collaboration. Healthcare systems globally face growing pressures:rising costs, workforce shortages and persistentinefficiencies. In this context, AI offers transformativeopportunities to enhance patient outcomes andoptimize system performance. However, realizingAI’s benefits in healthcare demands responsibledevelopment, rigorous evaluation and a deliberatefocus on building trust among stakeholders. –Independent quality assurance resourcesand real-world testing environments, suchas those being developed under initiativeslike the Testing and Experimentation Facilityfor Health AI and Robotics (TEF-Health), cansupport more dynamic development. 3.Promote public–private collaboration Today’s medicine regulatory frameworks – largelydesigned for pharmaceuticals and medicaldevices – are not fully suited to manage theprobabilistic, dynamic nature of AI technologies.Traditional evaluation methods, which emphas