“AI+医疗”应用边界引关注 专家建议全链条动态监管

Core Viewpoint - The core discussion revolves around the application boundaries of "AI + healthcare," emphasizing the need for AI to be an "assistant" rather than a replacement for healthcare professionals, while also addressing regulatory gaps and establishing a dynamic cross-departmental regulatory mechanism [1][2][3]. Group 1: Application Boundaries and Concerns - The refusal to integrate AI into hospital electronic medical record systems highlights concerns about young doctors becoming overly reliant on AI, potentially hindering their clinical thinking development [1]. - Experts argue that AI should enhance patient care without compromising the growth of medical professionals, suggesting a shift in perspective where AI serves to remind and verify clinical thinking rather than replace it [1][2]. - The debate centers on three main issues: prioritizing the cultivation of doctors' skills versus ensuring patient benefits, defining AI's role as a tool versus a potential decision-maker, and managing risks through either preemptive regulation or post-implementation adjustments [2]. Group 2: AI's Role in Healthcare - AI is increasingly being integrated into various clinical scenarios, from diagnostic imaging to intelligent pre-consultation and monitoring, with a significant rise in compliance and clinical penetration [3]. - By December 2025, 207 AI medical devices are expected to receive Class III medical device registration, indicating a maturation of AI in fields like biopharmaceuticals and diagnostic assistance [3]. - The essence of "AI + healthcare" is to optimize medical service processes and address resource shortages, reinforcing that AI is a supplement rather than a substitute for healthcare professionals [3][4]. Group 3: Risk and Ethical Challenges - The global market for AI healthcare solutions is projected to grow from $13.7 billion in 2022 to $155.3 billion by 2030, with China's market expected to reach $16.83 billion, indicating explosive growth opportunities [6]. - However, challenges such as commercialization, ethical dilemmas, and regulatory risks persist, particularly concerning data privacy, algorithm biases, and the opaque nature of AI decision-making [7][8]. - Ethical issues include patient awareness of AI involvement in their care, algorithm fairness, and the potential widening of the healthcare resource gap due to high costs of AI products [8]. Group 4: Regulatory Framework and Recommendations - Current regulations identify healthcare institutions and professionals as responsible parties, with AI positioned as an auxiliary tool, but lack detailed, dynamic control mechanisms for AI's unique challenges [10]. - Recommendations include establishing a risk classification system for AI medical products, creating an algorithm registration and review mechanism, and developing a risk warning platform to monitor AI applications in real-time [11]. - A regulatory sandbox mechanism is suggested for innovative AI healthcare products, allowing controlled exploration while adapting regulations to keep pace with technological advancements [11].