Research map / Probabilistic and scientific modeling / Molecules and proteins
Component-Wise Composite Likelihood Distillation for Censored Time-to-Event Data
Feiyang Deng, Lingfeng Luo, Jiayu Zhou, Kevin He
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 component-wise KL distillation from teacher | student · Simulated NCC test set | C-index: 0.550 -> 0.681 | Improved |
| add component-wise KL distillation from teacher | student · Simulated NCC test set | predictive deviance: 0.742 -> 0.688 | Improved |
| scale teacher score by 0.5 before distillation (ablation) | component-wise KL student · Simulated NCC test set | C-index: 0.681 -> 0.677 | No clear effect |
| shift teacher score by +2 before distillation (ablation) | component-wise KL student · Simulated NCC test set | C-index: 0.681 -> 0.678 | No clear effect |
| apply exponential monotone transform to teacher score (ablation) | component-wise KL student · Simulated NCC test set | C-index: 0.681 -> 0.635 | Hurt |
| apply exponential monotone transform to teacher score (ablation) | component-wise KL student · Simulated NCC test set | MeanAUC: 0.692 -> 0.640 | Hurt |
| exponential-transformed teacher still distilled vs none (ablation) | student · Simulated NCC test set | C-index: 0.550 -> 0.635 | Improved |
| use a good-quality instead of poor-quality teacher (more covariates) (ablation) | component-wise KL student · Simulated NCC | C-index, MeanAUC, deviance | Improved |
| use severely shifted instead of moderately shifted teacher population (ablation) | component-wise KL student · Simulated NCC covariate shift | test metrics | Hurt |
| add component-wise KL distillation from KI teacher | KP student · OPTN KP test cohort | C-index, MeanAUC, IBS, deviance | Improved |
| component-wise KL instead of MSE distillation | DeepNCC student · OPTN KI->KP donor-matched | stratified C-index: 0.5818 -> 0.5875 | Improved |
| component-wise KL instead of MSE distillation | MLP DeepSurv student · OPTN KI->KP donor-matched | stratified C-index: 0.5533 -> 0.5663 | Improved |
Closest papers by meaning
- Orthogonal Survival Learners for Estimating Heterogeneous Treatment Effects from Time-to-Event Data (NeurIPS 2025 · similarity 0.88)
- Heterogeneous Treatment Effect in Time-to-Event Outcomes: Harnessing Censored Data with Recursively Imputed Trees (ICML 2025 · similarity 0.87)
- Censor Dependent Variational Inference (ICML 2025 · similarity 0.87)
- Assessing the robustness of heterogeneous treatment effects in survival analysis under informative censoring (NeurIPS 2026 · similarity 0.87)
- SurvivalPFN: Amortizing Survival Prediction via In-Context Bayesian Inference (NeurIPS 2026 · similarity 0.87)
- Deep Copula-Based Survival Analysis for Dependent Censoring with Identifiability Guarantees (AAAI 2024 · 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.
