Eckhoff Jennifer A, Krauss Dolores T, Brunner Stefanie, Bruns Christiane J, Fuchs Hans F
Klinik für Allgemein‑, Viszeral‑, Thorax- und Transplantationschirurgie, Uniklinik Köln, Kerpener Straße 62, 50937, Köln, Deutschland.
Chirurgie (Heidelb). 2025 Sep 12. doi: 10.1007/s00104-025-02377-x.
Despite intensive research the clinical implementation of artificial intelligence (AI) in surgery remains limited. In addition to infrastructural and regulatory barriers, this is due to the isolated processing of individual data sources (e.g. video), although the true potential of surgical AI lies in the integration of multimodal data.
What added value does AI-driven analysis of multimodal data offer in surgery, and how can it realistically be integrated into clinical practice?
This review is based on first results on multimodal data acquisition and processing at University Hospital Cologne as well as a targeted literature search.
The integration of different data sources shows great potential; however, lack of infrastructure and regulation hinders the implementation.
In addition to technological development, clear legal frameworks are required to enable the clinical integration of innovative AI systems.
尽管进行了深入研究，但人工智能（AI）在外科手术中的临床应用仍然有限。除了基础设施和监管障碍外，这是由于对单个数据源（如视频）进行孤立处理，尽管外科人工智能的真正潜力在于多模态数据的整合。
人工智能驱动的多模态数据分析在外科手术中能提供哪些附加价值，以及如何切实地将其整合到临床实践中？
本综述基于科隆大学医院在多模态数据采集和处理方面的初步结果以及有针对性的文献检索。
不同数据源的整合显示出巨大潜力；然而，基础设施和监管的缺乏阻碍了其实施。
除了技术发展外，还需要明确的法律框架，以实现创新人工智能系统的临床整合。
