"Update on Reinforcement Learning-Based Blood Product Resuscitation in Gastrointestinal Hemorrhage"

We will provide an update our reinforcement learning-based algorithm to guide component blood product resuscitation for ICU patients with GI hemorrhage. We will provide evidence that a more liberal and balanced resuscitation strategy has better clinical outcomes, and we will provide an overview of ongoing areas of investigation relating to clinical state space representation strategies, reward structure design, and natural language processing integration in reinforcement learning.