Applying Reinforcement Learning to Epidemic Management: Strategic Influenza Control in Multiple Scenarios

This study addresses the challenge of controlling seasonal influenza, which causes significant global health burdens, by developing a Reinforcement Learning (RL)-based policy support system. Utilizing influenza data from Mainland China, our system optimizes non-pharmaceutical interventions (NPls) to balance health risks and economic activity across different epidemic scenarios. Our RL-based approach outperforms existing models in efficacy, demonstrating the system's potential in epidemic management. Additionally, we introduce the Variance-weighted Replay Memory Buffer (VW-RPM) optimization technique, enhancing the RL framework with a 2.76^% improvement in overall performance. This research advances RL applications in public health policy and epidemic management.