AI在健康行业赢得信任_21页_3mb

Summary of "Earning Trust for AI in Health: A Collaborative Path Forward"

Core Content

This white paper, published by the World Economic Forum in June 2025 in collaboration with Boston Consulting Group (BCG), outlines the critical steps necessary to build trust in artificial intelligence (AI) within the healthcare sector. It emphasizes the need for a collaborative, adaptive and globally coordinated approach to AI regulation and innovation.

Main Viewpoints

1. Urgent Need for Collaboration

• Global Divergences: Regulatory frameworks for AI are still evolving and vary significantly across regions. The Global North has more developed and standardized regulations, while the Global South faces fragmentation and under-resourcing, which can hinder AI adoption.
• Private Sector Leadership: The private sector is essential in driving innovation and standardization, especially in developing high-quality AI systems that can operate under diverse global regulations.
• Regulatory Evolution: Traditional regulatory models, designed for static products like pharmaceuticals, are not suitable for AI technologies, which evolve dynamically after deployment. Regulatory models must adapt to include lifecycle evaluation, post-market monitoring, and flexible frameworks.

2. Pragmatic Regulatory Approach

• Legislation as a Baseline: Governments are working on AI-related legislation, particularly in data protection and safety. However, the pace of legislative development is often slow and difficult to adapt.
• Regulatory Sandboxes: These testing environments allow for controlled innovation and are crucial in bridging the gap between regulation and technological advancement. They provide a safe space for private sector actors to test AI solutions while ensuring patient safety and privacy.
• Post-Market Surveillance: Ongoing monitoring of AI technologies is necessary to detect new risks and ensure continued safety and effectiveness. This includes real-world data analysis and iterative adaptation of AI systems.

3. Public-Private Partnerships (PPPs)

• Co-Development of Standards: PPPs are vital in developing and operationalizing regulatory standards for AI in healthcare. They enable the sharing of expertise, resources and risks to accelerate innovation.
• Examples of PPPs: Initiatives such as the Coalition for Health AI (CHAI) and the Trustworthy and Responsible AI Network (TRAIN) are examples of collaborative efforts that aim to harmonize AI standards and promote ethical use.
• Capacity Building: Enhancing technical literacy and regulatory capacity, especially in under-resourced health systems, is critical for global AI deployment and trust-building.

Key Information

• Healthcare Challenges: Rising costs, workforce shortages, and inefficiencies are pressing issues that AI can help address.
• AI in Healthcare: AI technologies, especially probabilistic models like generative AI (GenAI), introduce new risks and require different regulatory approaches compared to deterministic systems.
• Global Coordination: Harmonizing regulatory approaches across regions is necessary to enable scalable AI deployment and ensure equitable access to innovation.
• Regulatory Sandboxes: These environments provide flexibility for testing AI solutions while maintaining safety and ethical standards. Examples include the UK's AI Airlock.
• Post-Market Surveillance: This process is essential for tracking AI performance and adapting to new risks, ensuring that AI remains safe and effective over time.

Conclusion

The future of AI in healthcare depends on a balance between innovation and regulation. By strengthening evaluation processes, building technical capacity and fostering structured public-private collaboration, health systems can unlock the transformative potential of AI while upholding patient safety, trust and equitable access to care. The paper calls for immediate action to create a globally coordinated and adaptable regulatory environment that supports the responsible and effective deployment of AI technologies in healthcare.

Appendix Highlights

• Regulatory Sandbox Initiatives: The paper lists several regulatory sandbox initiatives in OECD and non-OECD countries, emphasizing their role in enabling innovation and ensuring safety.
• Global HealthAI Network: The Global Agency for Responsible AI in Health (HealthAI) is highlighted as a key player in supporting regulatory capacity, especially in the Global South.

Contributors and Acknowledgements

• The paper is a collaborative effort between the World Economic Forum and BCG, involving over 50 experts in the field of health AI.
• It acknowledges the importance of stakeholder input and the need for continued international cooperation in shaping the future of AI in healthcare.