GraphITE: Estimating Individual Effects of Graph-structured Treatments

Abstract

Outcome estimation of treatments for individual targets is a crucial foundation for decision making based on causal relations. Most of the existing outcome estimation methods deal with binary or multiple-choice treatments; however, in some applications, the number of interventions can be very large, while the treatments themselves have rich information. In this study, we consider one important instance of such cases, that is, the outcome estimation problem of graph-structured treatments such as drugs. Due to the large number of possible interventions, the counterfactual nature of observational data, which appears in conventional treatment effect estimation, becomes a more serious issue in this problem. Our proposed method GraphITE (pronounced 'graphite') obtains the representations of the graph-structured treatments using graph neural networks, and also mitigates the observation biases by using the HSIC regularization that increases the independence of the representations of the targets and the treatments. In contrast with the existing methods, which cannot deal with "zero-shot" treatments that are not included in observational data, GraphITE can efficiently handle them thanks to its capability of incorporating graph-structured treatments. The experiments using the two real-world datasets show GraphITE outperforms baselines especially in cases with a large number of treatments.

Learning causality with graphs

“…Individual treatment effect estimation with graph‐structured treatments . GraphITE (Harada and Kashima 2020) is a method, which addresses the problem of causal effect estimation with graph‐structured treatments. GraphITE learns representations of graph‐structured treatments with GNNs, and mitigates the estimation biases with a Hilbert–Schmidt independence criterion (HSIC) (Gretton et al.…”

A Survey of Deep Causal Models and Their Industrial Applications

“…In order to estimate impact of multi-cause treatment, Singlecause Perturbation [101] was proposed, which őrstly augments the observational dataset with potential outcomes estimated from single-cause interventions, and then adjusts the covariate of the augmented dataset to learn the estimator. Following that, GraphITE [142] was proposed, which utilizes graph neural networks to learn the representation of the graph-structured treatments. In order to reduce the observation biases, HSIC regularization is employed to obtain an independent representation of the targets in regard to the treatments.…”