Improving Sepsis Treatment Strategies by Combining Deep and Kernel-Based Reinforcement Learning

Sepsis is the leading cause of mortality in the ICU. It is challenging to manage because individual patients respond differently to treatment. Thus, tailoring treatment to the individual patient is essential for the best outcomes. In this paper, we take steps toward this goal by applying a mixture-o...

Full description

Saved in:
Bibliographic Details
Published inAMIA ... Annual Symposium proceedings Vol. 2018; pp. 887 - 896
Main Authors Peng, Xuefeng, Ding, Yi, Wihl, David, Gottesman, Omer, Komorowski, Matthieu, Lehman, Li-Wei H, Ross, Andrew, Faisal, Aldo, Doshi-Velez, Finale
Format Journal Article
LanguageEnglish
Published United States American Medical Informatics Association 2018
Subjects
Deep Learning
Fluid Therapy - adverse effects
Humans
Infusions, Intravenous
Intensive Care Units
Machine Learning
Medical History Taking
Observation
Retrospective Studies
Sepsis - therapy
Vasoconstrictor Agents - adverse effects
Vasoconstrictor Agents - therapeutic use
Online AccessGet full text

Cover

More Information
Summary:Sepsis is the leading cause of mortality in the ICU. It is challenging to manage because individual patients respond differently to treatment. Thus, tailoring treatment to the individual patient is essential for the best outcomes. In this paper, we take steps toward this goal by applying a mixture-of-experts framework to personalize sepsis treatment. The mixture model selectively alternates between neighbor-based (kernel) and deep reinforcement learning (DRL) experts depending on patient's current history. On a large retrospective cohort, this mixture-based approach outperforms physician, kernel only, and DRL-only experts.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:1942-597X
1559-4076