Research map / Probabilistic and scientific modeling / Treatment effects
DeepBlip: Estimating Conditional Average Treatment Effects Over Time
Haorui Ma, Dennis Frauen, Stefan Feuerriegel
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 |
|---|---|---|---|
| use DeepBlip instead of IGC-Net at long horizon | DeepBlip · MIMIC-III semi-synthetic (tau=10) | normalized RMSE: 1.50 -> 0.98 | Improved |
| use DeepBlip instead of Dynamic DML | DeepBlip · MIMIC-III semi-synthetic (tau=10) | normalized RMSE: 2.46 -> 0.98 | Improved |
| use double optimization trick instead of sequential training | Dynamic DML vs DeepBlip · MIMIC-III semi-synthetic | second-stage training time | Improved |
| add double optimization trick (detach future blip predictions) (ablation) | DeepBlip · MIMIC-III runtime | training time per epoch: 3.1 -> 3.6 min | Hurt |
| use DeepBlip under strong time-varying confounding | DeepBlip · Tumor growth (tau=2) | normalized RMSE | Improved |
| use G-computation methods (G-Net, IGC-Net) at low confounding | G-Net, IGC-Net · Tumor growth (tau=2) | normalized RMSE | Mixed |
| use balanced representations (CT, CRN) | CT, CRN · Tumor growth (tau=2) | normalized RMSE | No clear effect |
| use inverse propensity weighting (R-MSNs) | R-MSNs · Tumor growth and MIMIC-III | normalized RMSE | Hurt |
| swap transformer backbone for LSTM (ablation) | DeepBlip · Appendix E.1 ablation | CATE RMSE | No clear effect |
| treat vasopressor only in the last three hours vs first three | DeepBlip · MIMIC-III real-world | cumulative CATE on MAP: > 95% vs < 3% of total effect | Mixed |
Closest papers by meaning
- Continuous Treatment Effect Estimation Using Gradient Interpolation and Kernel Smoothing (AAAI 2024 · similarity 0.84)
- Orthogonal Survival Learners for Estimating Heterogeneous Treatment Effects from Time-to-Event Data (NeurIPS 2025 · similarity 0.84)
- Longitudinal Targeted Minimum Loss-based Estimation with Temporal-Difference Heterogeneous Transformer (ICML 2024 · similarity 0.84)
- Model-agnostic meta-learners for estimating heterogeneous treatment effects over time (ICLR 2025 · similarity 0.84)
- PairNet: Training with Observed Pairs to Estimate Individual Treatment Effect (ICML 2024 · similarity 0.84)
- Doubly robust identification of treatment effects from multiple environments (ICLR 2025 · similarity 0.84)
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.
