Reinforcement Learning, an artificial intelligence approach, has the potential to guide physicians in designing sequential treatment strategies for better patient outcomes but requires significant improvements before it can be applied in clinical settings, finds a new study.