Identifiability and Transportability in Dynamic Causal Networks

TL;DR

This paper introduces Dynamic Causal Networks as a causal extension of Dynamic Bayesian Networks, providing algorithms for effect identification and transportability across domains using passive observations.

Contribution

It presents a sound and complete algorithm for identifying causal effects in Dynamic Causal Networks and introduces a method for transporting causal effects between domains.

Findings

Effective algorithms for causal effect identification in dynamic settings.

Identification of two types of confounders affecting causal inference.

Method for transferring causal knowledge across different domains.

Abstract

In this paper we propose a causal analog to the purely observational Dynamic Bayesian Networks, which we call Dynamic Causal Networks. We provide a sound and complete algorithm for identification of Dynamic Causal Net- works, namely, for computing the effect of an intervention or experiment, based on passive observations only, whenever possible. We note the existence of two types of confounder variables that affect in substantially different ways the iden- tification procedures, a distinction with no analog in either Dynamic Bayesian Networks or standard causal graphs. We further propose a procedure for the transportability of causal effects in Dynamic Causal Network settings, where the re- sult of causal experiments in a source domain may be used for the identification of causal effects in a target domain.