New method learns hidden factors to improve treatment effect estimates
Proximal Causal Learning under Unmeasured Confounding
Machine Learning
Summary
Estimating how treatments affect people from data can be hard because important hidden factors may confuse the results. The paper presents a method that automatically finds special clues, called proxy variables, hidden within the data instead of needing experts to pick them. This helps get better estimates of treatment effects even when some confounding causes are not directly observed. The authors show their method works well on benchmark tasks and stays reliable under various difficult conditions.
What this means in practice
- •For healthcare data teams: Improve treatment effect estimates when patient data has hidden confounders using automatically learned proxy variables.
- •For policy analysts: Generate more reliable causal estimates from observational data involving unobserved factors by learning proxies rather than relying on expert-specified variables.
Authors
Ying Tang, Yi Wang
Abstract
Estimating treatment effects from observational data typically relies on the No Unmeasured Confounding Assumption (NUCA), which rarely holds in practice. Proximal causal learning (PCL) addresses unmeasured confounding via proxy variables, yet existing methods require the proxy variables to be pre-specified. Thus, we propose PCL-U, a framework that learns proxy variables directly from observed covariates. PCL-U uses neural encoders to decompose covariates into treatment-inducing, outcome-inducing, and shared proxies, guided by minimax mutual information objectives, and obtains causal estimates through a practical moment-based risk function. Experiments on benchmarks show that PCL-U matches or outperforms existing baselines. Besides, there are two types of synthetic datasets with varying dimensions and confounding strengths that illustrate that our method maintains stable estimation accuracy.