Reinforcement Learning for Clinical Decision Making

Catalog description

This course introduces the theory and application of Reinforcement Learning (RL) for improving clinical decision making in healthcare. Students will learn how sequential decision processes can be modeled using Markov Decision Processes (MDPs) and solved using modern RL algorithms such as dynamic programming, temporal-difference learning, Q-learning, and policy gradient methods. The course emphasizes challenges unique to healthcare limited data, partial observability, delayed outcomes, safety, fairness, interpretability, and clinician-in-the-loop design. Through a combination of lectures, hands-on programming, and case studies, students will analyze observational clinical datasets, build RL models for treatment optimization, and critically evaluate the reliability and ethical implications of deploying RL in real-world clinical settings. By the end of the course, students will be able to design, implement, and assess RL-based decision support tools tailored for medical applications.