Learning Optimal Dynamic Treatment Regimes Using Causal Tree Methods in Medicine

Abstract

Dynamic treatment regimes (DTRs) are used in medicine to tailor sequential treatment decisions to patients by considering patient heterogeneity. Common methods for learning optimal DTRs, however, have shortcomings: they are typically based on outcome prediction and not treatment effect estimation, or they use linear models that are restrictive for patient data from modern electronic health records. To address these shortcomings, we develop two novel methods for learning optimal DTRs that effectively handle complex patient data. We call our methods DTR-CT and DTR-CF. Our methods are based on a datadriven estimation of heterogeneous treatment effects using causal tree methods, specifically causal trees and causal forests, that learn non-linear relationships, control for time-varying confounding, are doubly robust, and explainable. To the best of our knowledge, our paper is the first that adapts causal tree methods for learning optimal DTRs. We evaluate our proposed methods using synthetic data and then apply them to real-world data from intensive care units. Our methods outperform state-of-the-art baselines in terms of cumulative regret and percentage of optimal decisions by a considerable margin. Our work improves treatment recommendations from electronic health record and is thus of direct relevance for personalized medicine.

Learning and Assessing Optimal Dynamic Treatment Regimes Through Cooperative Imitation Learning

“…The advantage of information extraction may be limited in case of less additional informative content available in the clinical free-text. Two methods using causal forests and causal trees and are based on a data-driven estimation of heterogeneous treatment effects are presented in [27]. These methods learn non-linear relationships and control for time-varying confounding.…”

Estimating optimal treatment regimes in survival contexts using an instrumental variable

“…Zhang et al (2012) proposed inverse propensity score weighted and augmented inverse propensity score weighted estimators for the normal outcome and Zhao et al (2015a) extended the results to time-to-event data. In recent years, more developments and studies along this research line have been published, referring to Qi and Liu (2018); Zhang and Zhang (2018); Pan and Zhao (2021), Blümlein et al (2022) for more details.…”