Research map / Probabilistic and scientific modeling / Causal discovery
Time Series Deconfounder: Estimating Treatment Effects over Time in the Presence of Hidden Confounders
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add inferred latent confounders (D_Z = 1) to MSM | MSM · MIMIC III, white blood cell count | RMSE × 10^2: 3.90 ± 0.00 -> 3.55 ± 0.05 | Helped |
| Add inferred latent confounders (D_Z = 1) to R-MSN | R-MSN · MIMIC III, blood pressure | RMSE × 10^2: 10.29 ± 0.05 -> 9.35 ± 0.11 | Helped |
| Add inferred latent confounders (D_Z = 1) to MSM | MSM · MIMIC III, oxygen saturation | RMSE × 10^2: 2.92 ± 0.00 -> 2.42 ± 0.02 | Helped |
| Add inferred latent confounders (D_Z = 1) to MSM | MSM · Tumor growth simulation, tumor volume | RMSE × 10^2: 7.29 ± 0.14 -> 6.47 ± 0.16 | Helped |
| Add inferred latent confounders (D_Z = 10) to R-MSN | R-MSN · Tumor growth simulation, tumor volume | RMSE × 10^2: 5.31 ± 0.16 -> 4.54 ± 0.17 | Helped |
| Use an MLP factor model instead of an RNN to infer latent confounders (ablation) | Time Series Deconfounder factor model · Synthetic data, treatment-assignment predictive checks over time | Predictive p-value closeness to 0.5 | Hurt |
| Remove multitask output from the RNN factor model (ablation) | Time Series Deconfounder factor model · Synthetic data, treatment-assignment predictive checks | Predictive p-value closeness to 0.5 | No clear effect |
| Increase inferred confounder dimension from D_Z = 1 to D_Z = 5 (ablation) | Time Series Deconfounder · Synthetic one-step-ahead treatment response estimation | RMSE | Mixed |
| Remove the observed single-cause confounder X_1 (ablation) | Time Series Deconfounder · Synthetic one-step-ahead treatment response estimation | RMSE | Hurt |
| Add inferred latent confounders when there are no hidden confounders (ablation) | Time Series Deconfounder · Synthetic one-step-ahead treatment response estimation, γ = 0 | RMSE | No clear effect |
Closest papers by meaning
- Causal Effect Inference with Deep Latent-Variable Models (NeurIPS 2017 · similarity 0.90)
- Instrumental Variable Estimation for Causal Inference in Longitudinal Data with Time-Dependent Latent Confounders (AAAI 2024 · similarity 0.90)
- Doubly robust identification of treatment effects from multiple environments (ICLR 2025 · similarity 0.89)
- LLM-Driven Treatment Effect Estimation Under Inference Time Text Confounding (NeurIPS 2025 · similarity 0.88)
- An Empirical Examination of Balancing Strategy for Counterfactual Estimation on Time Series (ICML 2024 · similarity 0.87)
- Causal Identification for Complex Functional Longitudinal Studies (ICLR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.