Research on epidemic tracking method based on reinforcement learning

Epidemiological investigation is the most critical measure for detecting and controlling risk factors, controlling the source of infection, and blocking the transmission routes in public health emergencies such as the COVID-19 pandemic. However, current epidemiological investigations are hindered by difficulties such as complex on-site conditions, diverse investigation issues, extensive and massive information, inefficient and repetitive investigations, and a reliance on manual analysis for tracing investigations, seriously affecting the efficiency and accuracy of epidemiological investigations. In this paper, artificial intelligence technology is used to automate and intelligentize epidemiological investigation and improve the efficiency and accuracy of epidemiological investigation, so as to help decision-making on epidemic prevention and control measures. For the close contact tracking problem, the deep reinforcement learning method is adopted. The agent learns how to take action to maximize the long-term reward by observing the changes of reward and state, so as to obtain the maximum possible answer through the input query.