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

Researchers propose a real-time surgical team dynamics model using time-expanded interaction graphs, where team members become time-indexed nodes and communication exchanges form directed edges, enabling static GNN inference on spatio-temporal data. The system predicts procedural efficiency as deviation from expected duration, with counterfactual analysis identifying minimal communication structure changes to improve outcomes. OR teams and surgical AI developers should take note.