Actionable Real-Time Modeling of Surgical Team Dynamics via Time-Expanded Interaction Graphs

The paper presents a novel approach to modeling surgical team dynamics through time-expanded interaction graphs, addressing the limitations of existing surgical AI systems that primarily focus on visual workflows. By representing team members as time-indexed nodes and their communications as directed edges, this method enables real-time analysis of intraoperative interactions, potentially improving procedural efficiency. The use of a static graph neural network for inference further enhances the model's applicability in surgical settings.