Disentangled Representation for Causal Mediation Analysis

TL;DR

This paper introduces DMAVAE, a deep learning model that disentangles different types of latent confounders to improve causal mediation analysis without relying on the sequential ignorability assumption.

Contribution

The work presents a novel disentangled variational autoencoder that accurately estimates causal effects by separating confounders into three types, overcoming limitations of existing methods.

Findings

Outperforms existing methods in causal effect estimation

Demonstrates strong generalisation in experiments

Shows potential for real-world applications

Abstract

Estimating direct and indirect causal effects from observational data is crucial to understanding the causal mechanisms and predicting the behaviour under different interventions. Causal mediation analysis is a method that is often used to reveal direct and indirect effects. Deep learning shows promise in mediation analysis, but the current methods only assume latent confounders that affect treatment, mediator and outcome simultaneously, and fail to identify different types of latent confounders (e.g., confounders that only affect the mediator or outcome). Furthermore, current methods are based on the sequential ignorability assumption, which is not feasible for dealing with multiple types of latent confounders. This work aims to circumvent the sequential ignorability assumption and applies the piecemeal deconfounding assumption as an alternative. We propose the Disentangled Mediation Analysis Variational AutoEncoder (DMAVAE), which disentangles the representations of latent confounders into three types to accurately estimate the natural direct effect, natural indirect effect and total effect. Experimental results show that the proposed method outperforms existing methods and has strong generalisation ability. We further apply the method to a real-world dataset to show its potential application.