Research map / Probabilistic and scientific modeling / Treatment effects
Smooth Multi-Policy Causal Effect Estimation in Longitudinal Settings
Wenxin Chen, Weishen Pan, Kyra Gan, Fei Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 policy-aware shared encoder (PEQ-Net) over DeepLTMLE | DeepLTMLE · Limited confounding, deterministic CF 3a | RMSE: 1.33 -> 0.27 | Improved |
| add policy-aware shared encoder (PEQ-Net) over DeepLTMLE | DeepLTMLE · Expanded confounding, deterministic CF 2a | RMSE: 1.42 -> 0.25 | Improved |
| add policy-aware shared encoder (PEQ-Net) over DeepLTMLE | DeepLTMLE · Limited DGP, dynamic policies differ in first two steps, CF 1b | RMSE: 0.1255 -> 0.0026 | Improved |
| add policy-aware shared encoder (PEQ-Net) over DeepLTMLE | DeepLTMLE · Expanded DGP, dynamic policies differ in first two steps, CF 1b | RMSE: 0.1302 -> 0.0003 | Improved |
| add policy-aware shared encoder (PEQ-Net) over DeepLTMLE | DeepLTMLE · Limited DGP, dynamic policies differ in all steps, CF 1c | RMSE: 0.2725 -> 0.1220 | Improved |
| add policy-aware shared encoder (PEQ-Net) over DeepLTMLE | DeepLTMLE · Expanded DGP, dynamic policies differ in all steps, CF 2c | RMSE: 0.4956 -> 0.2715 | Improved |
| remove LTMLE targeting step, evaluate ICE-only Q for both models (ablation) | PEQ-Net vs DeepLTMLE · MIMIC-III semi-synthetic | CATE error | Improved |
| share parameters via independent fine-tuning per policy instead of fully separate models (ablation) | DeepLTMLE · MIMIC-III semi-synthetic | RMSE | Improved |
| use multi-Q-head shared backbone rather than independent fine-tuning (ablation) | DeepLTMLE · MIMIC-III semi-synthetic | RMSE | Improved |
| use PEQ-Net policy embedding instead of multi-Q-head or fine-tuning (ablation) | DeepLTMLE with parameter sharing · MIMIC-III semi-synthetic | RMSE | Improved |
| add extreme policies 0 and 1 to joint training set (ablation) | PEQ-Net · Joint training, CF 1c | RMSE: 0.14 -> 0.18 | Hurt |
| add extreme policies 0 and 1 to joint training set (ablation) | PEQ-Net · Joint training, CF 2c | RMSE: 0.27 -> 0.14 | Improved |
Closest papers by meaning
- CausalPFN: Amortized Causal Effect Estimation via In-Context Learning (NeurIPS 2025 · similarity 0.87)
- Learning Joint Interventional Effects from Single-Variable Interventions in Additive Models (ICML 2025 · similarity 0.86)
- Coupling Generative Modeling and an Autoencoder with the Causal Bridge (NeurIPS 2025 · similarity 0.86)
- Causal Identification for Complex Functional Longitudinal Studies (ICLR 2025 · similarity 0.85)
- Towards Estimating Bounds on the Effect of Policies under Unobserved Confounding (NeurIPS 2024 · similarity 0.85)
- Learning Structural Causal Models from Ordering: Identifiable Flow Models (AAAI 2025 · similarity 0.85)
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.
