An AI Coach for Surgical Teamwork

Cardiac surgery is crucial for treating serious heart conditions, with over 900,000 procedures performed annually. In these high-stakes operations, teams comprising surgeons, anesthesiologists, perfusionists, and nurses work together in the complex environment of the cardiac Operating Room (OR). Ideal teamwork is essential for success, but this is often challenged by factors like high workload, fatigue, and interruptions, which can lead to human errors and impact patient safety. Our project aims to mitigate these risks by developing an AI-enabled coaching system (AI Coach) designed to enhance surgical teamwork in the cardiac OR.

The envisioned AI Coach system will combine data from multimodal sensors, novel data-driven algorithms, and an intuitive user interface, all aimed at enhancing teamwork in cardiac surgery settings. At the heart of the AI Coach system are innovative: multi-agent imitation learning algorithms to learn generative models of surgical teamwork from behavioral and physiological data; and explainable AI algorithms to generate interpretable feedback for improving surgical teamwork. In collaboration with medical and human factors experts, the project will refine this system by employing iterative, user-centered design approaches. Please see the recent research outcomes (listed below) to learn more.

This cross-disciplinary project is a collaboration with the Medical Robotics and Computer Assisted Surgery Lab of VA Boston, Interactive Robotics Group of MIT, the STRATUS Center for Medical Simulation of Harvard Medical School, and Dr. Eduardo Salas at Rice University. We acknowledge the NSF/NIH Smart Health and Biomedical Research in the Era of Artificial Intelligence and Advanced Data Science Program for supporting this project.


  1. AAMAS
    IDIL: Imitation Learning of Intent-Driven Expert Behavior
    In International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), 2024
  2. AAAI
    AI-Assisted Human Teamwork
    In AAAI Conference on Artificial Intelligence (AAAI) Doctoral Consortium, 2024
  3. A Novel Multimodal Perspective on Objective Assessment of Non-Technical Skills in Cardiac Surgery
    Mahdi Ebnali, Lauren Kennedy-Metz, Giovanna Varni, Vaibhav Unhelkar, Eduardo Salas, Roger Dias, Marco Zenati
    Extended Abstract at the Academic Surgical Congress (ASC), 2024
  4. Towards a Web-Based Digital Twin for the Cardiac Operating Room
    Arnan Adhikari, Sangwon Seo, Vaibhav Unhelkar
    Poster at the Ken Kennedy Institute AI in Health Conference (AIHC), 2023
    Using Deep Learning to Assess Teamwork during Cardiac Surgery
    Mahdi Ebnali, Marco Zenati, Steven Yule, Vaibhav Unhelkar, Roger Dias
    Extended Abstract at the Clinical Translation of Medical Image Computing and Computer Assisted Interventions (CLINICCAI), 2023
  6. Opportunities and Challenges of Real-Time Measurement of Team Performance in the Cardiac Operating Room
    Maha Khalid, Sangwon Seo, Marco Zenati, Mahdi Ebnali, Lauren Kennedy-Metz, Roger Dias, Vaibhav Unhelkar, Eduardo Salas
    Extended Abstract at the 67th International Annual Meeting of the Human Factors and Ergonomics Society (HFES), 2023
  7. AAMAS
    Automated Task-Time Interventions to Improve Teamwork using Imitation Learning
    In International Conference on Autonomous Agents and Multi-Agent Systems (AAMAS), 2023
  8. IJCAI
    Semi-Supervised Imitation Learning of Team Policies from Suboptimal Demonstrations
    In International Joint Conference on Artificial Intelligence (IJCAI), 2022
  9. CogSIMA
    Towards an AI Coach to Infer Team Mental Model Alignment in Healthcare
    Sangwon Seo, Lauren Kennedy-Metz, Marco Zenati, Julie Shah, Roger Dias, Vaibhav Unhelkar
    In International Conference on Cognitive and Computational Aspects of Situation Management (CogSIMA), 2021