Sequential Treatment Effect Estimation with Unmeasured Confounders

2025-05-14

AI Summary

Estimating cumulative causal effects remains challenging in sequential decision-making settings with unobserved confounding. To address this, we propose DSIV-CFR—a novel framework that, for the first time, leverages prior outcomes as negative control outcomes and jointly models them with latent instrumental variables (IVs) embedded in observed covariates. Under the negative control assumption, our approach enables implicit IV decomposition and establishes identifiability conditions in temporal settings. DSIV-CFR integrates Transformer-based sequential modeling, IV theory, generalized method of moments (GMM), and counterfactual regression to robustly correct bias induced by unmeasured confounding. Evaluated on four benchmark datasets, it achieves significant improvements in both one-step and multi-step causal effect estimation accuracy, thereby enabling more reliable learning of dynamic optimal treatment policies.

Technology Category

• Reasoning under Uncertainty: Sequential Decision Making
• Machine Learning: Causal Learning
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

This paper studies the cumulative causal effects of sequential treatments in the presence of unmeasured confounders. It is a critical issue in sequential decision-making scenarios where treatment decisions and outcomes dynamically evolve over time. Advanced causal methods apply transformer as a backbone to model such time sequences, which shows superiority in capturing long time dependence and periodic patterns via attention mechanism. However, even they control the observed confounding, these estimators still suffer from unmeasured confounders, which influence both treatment assignments and outcomes. How to adjust the latent confounding bias in sequential treatment effect estimation remains an open challenge. Therefore, we propose a novel Decomposing Sequential Instrumental Variable framework for CounterFactual Regression (DSIV-CFR), relying on a common negative control assumption. Specifically, an instrumental variable (IV) is a special negative control exposure, while the previous outcome serves as a negative control outcome. This allows us to recover the IVs latent in observation variables and estimate sequential treatment effects via a generalized moment condition. We conducted experiments on 4 datasets and achieved significant performance in one- and multi-step prediction, supported by which we can identify optimal treatments for dynamic systems.

Problem

• Estimating sequential treatment effects with unmeasured confounders
• Addressing latent confounding bias in dynamic decision-making
• Proposing DSIV-CFR framework for accurate effect estimation

Innovation

• Uses transformer to model time sequences
• Introduces DSIV-CFR for latent confounding
• Employs instrumental variables via negative controls