编者按     

   2025年10月24~26日，第十届国家癌症中心结直肠癌学术会议在北京隆重召开。会议聚集国内外权威学者，围绕结直肠癌诊疗领域的热点和前沿进展进行深入讨论。会上，梅奥诊所的Claude Deschamps教授应邀发表了题为《人工智能与肿瘤》的专题报告，系统阐述了人工智能（AI）在肿瘤诊疗中的临床应用前景。 

 Claude Deschamps教授在外科领域有着深厚的领导力和丰富的经验，为进一步分享其学术观点，《肿瘤瞭望消化时讯》记者在会后对教授进行了独家专访。Claude Deschamps教授针对AI在手术中如何协助外科医生进行实时决策、未来AI与临床医学如何深度融合、AI算法的临床验证以及AI未来应用前景等内容发表了深刻见解，以下内容是对本次专访核心要点的梳理与总结。      

              肿瘤瞭望消化时讯 

 在您看来，人工智能在实现数据分析的同时，如何在手术过程中直接协助外科医生进行复杂的实时决策？实现这一目标最大的技术和伦理障碍是什么？     

  Claude Deschamps教授：首先，在手术室场景下，AI已经能够为外科医生提供实时获取专业知识的可能性。我们拥有所谓的“知识墙”，即一块电子面板，可随着手术推进实时提供数据。这些数据包括患者的生命体征、病理检查结果，以及手术各步骤的相关信息——这便是AI在手术过程中辅助手术推进的核心方式。其次，AI能增强外科诊疗规范性，并非以僵化方式，而是具备灵活性的辅助。通过AI可使外科诊疗实现标准化，确保不同的外科医生、不同的医疗机构、不同的国家/地区在治疗同一疾病、同一分期患者时，采用统一的规范方案。此外，还需强调外科诊疗团队协作的重要性。AI有助于凝聚团队力量，推动“专家组成的团队”向“专家级团队”转变——此时团队本身即为核心专家，而非依赖个体。借助AI提升团队协作能力，可进一步优化患者治疗效果。 

 最后，我想要说的一点是，我们需要提升技术的可及性与可负担性，这是当前面临的最大技术与伦理障碍。我们不应该让AI成为少数群体的工具，也不是成为顶尖机构的工具，而是让AI技术惠及所有人。唯有如此，才能突破伦理与技术层面的双重障碍。当然，数据质量与数据完整性也至关重要。谈及伦理问题时，必须考虑数据完整性：数据是否真实可靠？数据处理过程是否遵循科学严谨性与伦理规范？ Prof Claude Deschamps：Frst, there is already the possibility in the operating room of providing the surgeon with real-time knowledge during the surgery. What we have is what we call a "wall of knowledge"—an electronic panel that provides data as the operation progresses. There is information about how the patient is doing, information about the results of the pathology, and information about the different steps of the operation. This is how AI can support the progress during an operation. Another aspect of AI that can strengthen surgical practice is standardization. AI can help with this not rigidly, but flexibly: it can bring standardization to surgical practice, ensuring that the same disease at the same stage is treated the same way across different surgeons, different institutions, and different countries.Finally, there is the importance of the team caring for the patient alongside the surgeon. AI helps unite the team: instead of a team of experts, you become an expert team. The team is the expert, not the individual. And by leveraging this ability to work as a team, AI can help treat patients better. The last thing I'd like to say is that we need to make technology more accessible and more affordable. What is the biggest technical and ethical hurdle here? It is not letting AI become a tool for the rich or a tool for the best institutions—it is making AI available to everyone. That is how we can overcome the ethical and technical hurdles.

 Finally, we must emphasize the importance of data quality and data integrity. When we talk about ethics, we have to think about integrity: Is the data accurate? Has the data been handled with respect and scientific rigor?     

        肿瘤瞭望消化时讯

  认为，在AI驱动的未来，人类外科医生不可替代的核心价值是什么？下一代外科医生应该重点培养哪些特定技能，以便与AI协同工作，而非被其替代？     

  Claude Deschamps教授：人类外科医生不可替代的核心价值在于同理心——即理解患者的能力，这本质上是一种人际连接。仅依赖技术工具，不倾注人文关怀、不主动深度思考、不将二者结合的医生，无法成为一名优秀的医生——这便是外科医生不可替代的核心价值。此外，特定技能，如适应性与应变能力等也至关重要。应将AI视为工具，而非人类的替代品。AI就如同手机，我与您交流时，手机不会取代我，但它能在诸多方面帮我提升沟通效率——AI的作用仅此而已。AI或许是一种复杂工具，但终究只是工具，无法取代人类。最后，我认为下一代外科医生关键特质在于勇于尝试，勇于创新。若想利用这些工具发展，这些便是不可或缺的核心技能。 Prof Claude Deschamps：The irreplaceable core value is empathy—the ability to understand your fellow human beings. That is human connection. One cannot be a good physician by relying solely on technical tools without using his or her heart, without engaging the mind, without connecting both. That is the irreplaceable core value. Specific skills, such as adaptability and the ability to adapt, are also crucial. Consider AI as a tool, not a replacement for the human. AI is like a cell phone; it doesn't replace me when I talk to you, but it helps me do a better job in many ways. That's all. AI may be a complex tool, but it is still a tool and does not replace the human. Finally, the last key attribute is a willingness to experiment and a willingness to innovate. These are the essential skills if we are to evolve alongside these tools.     

        肿瘤瞭望消化时讯

  要让AI算法在手术等高风险场景中被认可 ，需要严格的临床验证。在您看来，一个用于外科AI工具的“金标准”临床试验应具备哪些要素？如何构建全球框架，确保这些技术兼具有效性与公平性？     

  Claude Deschamps教授：临床验证的金标准仍是随机对照试验（RCT）。即便已有AI工具，也可通过RCT确保所生成数据的质量。此外，内部验证与外部验证均不可或缺：不能仅依赖内部验证，否则会导致数据无法实现广泛共享；也不能仅依赖外部验证，否则会在自身机构内部丧失可信度。最后，必须加大教育投入。教育对AI的广泛应用至关重要，具体而言，需完善高质量的继续医学教育（CME）项目，并建立专业能力维持体系。这些是确保AI医疗技术兼具有效性与公平性的关键举措。 Prof Claude Deschamps：The gold standard remains the randomized clinical trial (RCT). Even for AI tools, we can use RCTs to ensure the quality of the data generated. Additionally, both internal and external validation are necessary. You cannot rely solely on internal validation, as this will prevent you from sharing data widely; nor can you rely solely on external validation, as this will lack credibility within your own organization. Finally, we must invest in education. Education is critical for the widespread adoption of such technologies. Specifically, we need strong continuing medical education (CME) programs and systems to maintain professional competence. These are the key steps to ensuring AI medical technologies are both effective and equitable.     

        肿瘤瞭望消化时讯

  展望未来10年，您能否描绘AI将如何从根本上改变患者的诊疗历程——从初始诊断、手术规划到长期随访？您认为最令人期待的潜在可能性是什么？     

  Claude Deschamps教授：我小时候，家里的电话是固定在墙上的。根本无法想象，三四十年后，我们还会用同样的电话。AI也是如此，它本质上是一种工具：若能辅以完善的教育体系与广泛的可及性，AI不仅能为医疗健康领域带来益处，还能惠及生活的方方面面。 

 诚然，AI存在风险与局限性，我们必须制定相应的规范准则与防护机制。但我坚信，若能合理运用AI，它最终将让世界变得更美好。例如，我预计患者教育领域将发生显著变革：未来患者就诊时，其自身的健康知识储备将大幅提升。如果运用得当，AI将会显著提升医疗服务质量。     Prof Claude Deschamps：When I was a little boy, the telephone was a device attached to the wall at home. At that time, one could hardly imagine that thirty or forty years later, you would have that same thing. Similarly, Artificial Intelligence (AI) is a tool. With proper education and widespread availability, it will bring benefits not only to healthcare but to all areas of life. Admittedly, there are dangers and limitations; there are guidelines we must follow and guardrails we must put in place. However, I firmly believe that, if used wisely, AI will ultimately make our world a better place. For instance, I anticipate that patient education will be significantly enhanced. Patients will likely arrive at doctor's appointments with a substantially higher baseline of knowledge. This informed patient base, coupled with the prudent application of AI, can markedly improve the quality of care we provide.

0