Combining Reinforcement Learning with Supervised Learning for Sepsis Treatment

Sepsis is one of the leading causes of mortality globally that costs billions of dollars annually. Until now, the general method of treatment for sepsis remains uncertain. Therefore, treating septic patients is highly challenging. Some recent research has successfully applied reinforcement learning to generate optimal treatment policies for septic patients. The policies are proved to be better than that of physicians but sometimes they can suggest some actions that the clinicians almost never used. In this paper, we propose a method of combining supervised learning and reinforcement learning using Mixture-of-Experts technique. The policy derived from our model outperforms the physicians’ policies and limit the number of dangerous actions. It can be used as a dynamic decision-supporting tool for clinicians to reduce the mortality of patients.