Towards Reinforcement Learning for Surgical Decision-Making

Presenter: Dr. Tanner Wilson 

With an eye on small bowel obstruction as a problem of iterative surgical decision-making, Dr. Wilson will present a smorgasbord of machine learning methods to predict failure of non-operative management. HWe will illustrate the shortcomings of one-shot prediction methods and present reinforcement learning as a meaningful alternative. Finally, he will present the many questions and challenges facing an investigator attempting to implement reinforcement learning in a high-acuity clinical context.