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

TL;DR

This paper introduces a framework to identify key decision points in continuous health trajectories, enabling faster, interpretable reinforcement learning policies for hypotension treatment from batch data.

Contribution

It develops a novel method to compress continuous trajectories into key decision points, improving interpretability and planning efficiency in healthcare reinforcement learning.

Findings

Faster policy computation due to reduced state space

Enhanced interpretability for clinical experts

Effective application to hypotensive patient data

Abstract

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.