DeCaFlow: A deconfounding causal generative model
This addresses the challenge of causal inference in complex systems with hidden confounders for researchers and practitioners in fields like biology or social sciences, offering a generalizable tool, though it builds on prior work in causal estimation.
The paper tackles the problem of causal inference with hidden confounders using observational data and a causal graph, introducing DeCaFlow, which provides accurate estimates for all identifiable causal and counterfactual queries, outperforming existing methods on datasets like Ecoli70 with multiple confounders and hundreds of queries.
We introduce DeCaFlow, a deconfounding causal generative model. Training once per dataset using just observational data and the underlying causal graph, DeCaFlow enables accurate causal inference on continuous variables under the presence of hidden confounders. Specifically, we extend previous results on causal estimation under hidden confounding to show that a single instance of DeCaFlow provides correct estimates for all causal queries identifiable with do-calculus, leveraging proxy variables to adjust for the causal effects when do-calculus alone is insufficient. Moreover, we show that counterfactual queries are identifiable as long as their interventional counterparts are identifiable, and thus are also correctly estimated by DeCaFlow. Our empirical results on diverse settings (including the Ecoli70 dataset, with 3 independent hidden confounders, tens of observed variables and hundreds of causal queries) show that DeCaFlow outperforms existing approaches, while demonstrating its out-of-the-box applicability to any given causal graph. An implementation can be found in https://github.com/aalmodovares/DeCaFlow