Supervised reinforcement learning for recommending treatment strategies in sepsis

Haoran Sun; Qiuming Chen; Yujing Yang

doi:10.1117/12.2689783

10 August 2023 Supervised reinforcement learning for recommending treatment strategies in sepsis

Haoran Sun, Qiuming Chen, Yujing Yang

Proceedings Volume 12748, 5th International Conference on Information Science, Electrical, and Automation Engineering (ISEAE 2023); 127482Q (2023) https://doi.org/10.1117/12.2689783
Event: 5th International Conference on Information Science, Electrical and Automation Engineering (ISEAE 2023), 2023, Wuhan, China

Abstract

Sepsis is a complex syndrome that leads to shock or death in the intensive care unit (ICU). The influence of multiple factors combined with the patient’s heterogeneity makes treating sepsis challenging to personalize. Research in reinforcement learning (RL) has made significant progress in medical decision-making in recent years. In addition, combining deep learning with RL is gradually becoming a mainstream approach to solving complex sequential problems, which provides a solid theoretical foundation for our work. This work proposes a supervised reinforcement learning (SRL) model to recommend personalized fluid and boosting drug doses for sepsis patients. In this context, RL aims to maximize the expected return to find the optimal treatment strategy, and supervised learning minimizes the difference with physician decisions to reduce the occurrence of high-risk strategies. The two complement each other, allowing the model to learn reasonable treatment strategies to assist physicians in making medical decisions to improve treatment outcomes. The results show that the SRL model can discover more optimal treatment strategies to provide personalized interventions.

Citation Download Citation

Haoran Sun, Qiuming Chen, and Yujing Yang "Supervised reinforcement learning for recommending treatment strategies in sepsis", Proc. SPIE 12748, 5th International Conference on Information Science, Electrical, and Automation Engineering (ISEAE 2023), 127482Q (10 August 2023); https://doi.org/10.1117/12.2689783

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available