Sandy Engelhardt ,Georgii Kostiuchik ,Branislav Bezak ,Jacob Chacko ,Edgar Daeter ,Hazem Fallouh ,Philippe Grieshaber ,Nabil Hussein ,Alexander Meyer ,Paola Quattroni...Show more

Abstract

Artificial intelligence (AI) has made significant advancements across various surgical disciplines. Its evolution has been remarkable, transitioning from single-purpose models to sophisticated multi-task foundation models, including multimodal models based on transformers or AI agents including large language models (LLMs). Vision language models and Generative AI are currently transforming the way surgical procedures are analysed. These advancements have the potential to enhance decision-making, precision, and patient outcomes. However, one of the most pressing challenges in surgical AI is data integration. The complex and dynamic nature of the surgical environment makes it difficult to aggregate and utilize diverse data sources effectively. Ensuring seamless integration is crucial for optimizing AI applications in surgery. In this work, we highlight existing resources, key concepts, and notable research efforts in education and training and perioperative support. By showcasing examples from cardiothoracic surgery (CTS) and other fields, we aim to provide insights into AI’s expanding role and the ongoing efforts to enhance its effectiveness in surgical practice.