Research map / Probabilistic and scientific modeling / Treatment effects
SurvITE: Learning Heterogeneous Treatment Effects from Time-to-Event Data
Alicia Curth, Changhee Lee, Mihaela van der Schaar
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 IPM balancing term towards baseline population (ablation) | SurvITE · S4 (zeta=3, no overlap) | RMSE HTE_rmst L=20: 0.930 -> 0.703 | Improved |
| add IPM balancing term (ablation) | SurvITE · S3 (zeta=3, no overlap) | RMSE HTE_rmst L=10: 0.275 -> 0.225 | Improved |
| add IPM balancing term (ablation) | SurvITE · Twins (no censoring) | RMSE HTE_rmst L=180: 19.80 -> 18.34 | Improved |
| use baseline-population target instead of CFR-1 balancing at baseline only (ablation) | SurvITE · S4 (zeta=3, no overlap) | RMSE HTE_rmst L=20: 1.016 -> 0.703 | Improved |
| use baseline-population target instead of CFR-2 balancing at each time step (ablation) | SurvITE · S4 (zeta=3, no overlap) | RMSE HTE_rmst L=20: 0.815 -> 0.703 | Improved |
| CFR-2 vs full SurvITE in S3 (ablation) | SurvITE · S3 (zeta=3, no overlap) | RMSE HTE_rmst L=20: 0.691 vs 0.687 | No clear effect |
| CFR-1 balancing instead of no IPM in S4 (ablation) | SurvITE (no IPM) · S4 (zeta=3, no overlap) | RMSE HTE_rmst L=10: 0.310 -> 0.341 | Hurt |
| SurvITE architecture without IPM vs DeepHit | SurvITE (no IPM) · S3 (zeta=3, no overlap) | RMSE HTE_rmst L=20: 0.821 (DeepHit) vs 0.843 | Mixed |
| SurvITE vs DeepHit | SurvITE · S4 (zeta=3, no overlap) | RMSE HTE_rmst L=20: 0.830 -> 0.703 | Improved |
| SurvITE vs CSA generative model | SurvITE · Twins (censoring) | RMSE HTE_rmst L=180: 47.79 -> 18.76 | Improved |
| SurvITE vs Cox | SurvITE · Twins (no censoring) | RMSE HTE_rmst L=30: 2.85 -> 2.53 | Improved |
| SurvITE vs RSF on Twins | SurvITE · Twins (no censoring) | RMSE HTE_rmst L=180: 22.42 -> 18.34 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DeepHit — DeepHit: A Deep Learning Approach to Survival Analysis With Competing Risks (AAAI 2018) · better on 8 of 8 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Orthogonal Survival Learners for Estimating Heterogeneous Treatment Effects from Time-to-Event Data (NeurIPS 2025 · similarity 0.91)
- Assessing the robustness of heterogeneous treatment effects in survival analysis under informative censoring (NeurIPS 2026 · similarity 0.89)
- Heterogeneous Treatment Effect in Time-to-Event Outcomes: Harnessing Censored Data with Recursively Imputed Trees (ICML 2025 · similarity 0.88)
- Quantifying Aleatoric Uncertainty of the Treatment Effect: A Novel Orthogonal Learner (NeurIPS 2024 · similarity 0.86)
- Inferring the Long-Term Causal Effects of Long-Term Treatments from Short-Term Experiments (ICML 2024 · similarity 0.86)
- Model-agnostic meta-learners for estimating heterogeneous treatment effects over time (ICLR 2025 · similarity 0.86)
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.
