Reproduce “When Reconstruction Is Not Enough: Doubly Robust Counterfactual and Interventional Mediation Analysis with Learned Multimodal Representations” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Cong Cao
The authors examine whether compressed health-record representations preserve information needed for causal inference, especially mediation. They show that reconstruction and standard safeguards cannot recover discarded confounders, and propose checking representations against causal nuisance functions while retaining known confounders. In simulation, a dimension-four VAE had 0.015 coverage for nominal 95% intervals; adding prespecified covariates raised coverage to 0.965, in this design.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.