Counterfactually Guided Off-policy Transfer in Clinical Settings

TL;DR

This paper introduces a causal, counterfactual approach for off-policy transfer in healthcare, effectively addressing data scarcity and unobserved confounding to improve treatment policy learning in new patient populations.

Contribution

It presents a novel causal modeling method that leverages source domain priors and counterfactual trajectories for transfer learning under data scarcity and confounding.

Findings

Significant improvement in treatment policy performance on simulated sepsis task.

Effective handling of unobserved confounding in transfer learning.

Maintains stability and interpretability through causal parametrization.

Abstract

Domain shift, encountered when using a trained model for a new patient population, creates significant challenges for sequential decision making in healthcare since the target domain may be both data-scarce and confounded. In this paper, we propose a method for off-policy transfer by modeling the underlying generative process with a causal mechanism. We use informative priors from the source domain to augment counterfactual trajectories in the target in a principled manner. We demonstrate how this addresses data-scarcity in the presence of unobserved confounding. The causal parametrization of our sampling procedure guarantees that counterfactual quantities can be estimated from scarce observational target data, maintaining intuitive stability properties. Policy learning in the target domain is further regularized via the source policy through KL-divergence. Through evaluation on a simulated sepsis treatment task, our counterfactual policy transfer procedure significantly improves the performance of a learned treatment policy when assumptions of "no-unobserved confounding" are relaxed.