Reinforcement Learning for Adaptive Healthcare: Deep Q-Networks in Clinical Decision Optimization

Abstract

The chapter explores the conditioning of Deep Q Networks (DQNs) and computational intelligence for changing the healthcare decision making. We study the role of reinforcement learning for sequential optimization of clinical decisions in different kinds of healthcare domains. Innovations regarding application of DQNs in complex medical environments as well as the theoretical foundations of DQNs are presented. This paper shows how these technologies transform clinical decision supporting systems, predictive analytics, disease diagnosis, personalized treatment optimization as well as IoT-enabled patient monitoring. Demonstrating wide spread implementation in settings of chronic disease management, critical care and diagnostic pathways results in very substantial improvements of clinical outcomes over traditional approaches.