Identifying Decision Points for Safe and Interpretable Reinforcement Learning in Hypotension Treatment

Many batch RL health applications first discretize time into fixed intervals. However, this discretization both loses resolution and forces a policy computation at each (potentially fine) interval. In this work, we develop a novel framework to compress continuous trajectories into a few, interpretab...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2021-01
Main Authors: Zhang, Kristine, Wang, Yuanheng, Du, Jianzhun, Chu, Brian, Celi, Leo Anthony, Kindle, Ryan, Doshi-Velez, Finale
Format: Article
Language:English
Subjects:
Hypotension
Inspection
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Many batch RL health applications first discretize time into fixed intervals. However, this discretization both loses resolution and forces a policy computation at each (potentially fine) interval. In this work, we develop a novel framework to compress continuous trajectories into a few, interpretable decision points --places where the batch data support multiple alternatives. We apply our approach to create recommendations from a cohort of hypotensive patients dataset. Our reduced state space results in faster planning and allows easy inspection by a clinical expert.
ISSN:2331-8422