Research map / Label-efficient and robust learning / Semi-supervised learning
Surrogate Assisted Semi-supervised Inference for High Dimensional Risk Prediction
Jue Hou, Zijian Guo, Tianxi Cai
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 SAS (surrogate-assisted semi-supervised) instead of supervised LASSO | risk model · Simulation Scenario I (S(beta)=174, strong surrogate) | AUC: 0.660 -> 0.711 | Improved |
| use SAS instead of supervised LASSO | risk model · MGB Biobank T2DM | CV AUC: 0.488 -> 0.763 | Improved |
| use SAS instead of supervised LASSO | linear prediction x_new^L · Scenario I (moderate S(beta)) | rMSE: 0.719 -> 0.229 | Improved |
| use SAS instead of supervised LASSO | linear prediction, dense x_new · Scenario I (moderate S(beta)) | rMSE: 0.423 -> 0.141 | Improved |
| use strong instead of moderate surrogates (ablation) | SAS · Scenario I (moderate S(beta), x_new^L) | rMSE: 0.298 -> 0.229 | Improved |
| use strong instead of moderate surrogates (ablation) | SAS · Scenario II (moderate S(beta), x_new^S) | rMSE: 0.100 -> 0.072 | Improved |
| use strong instead of moderate surrogates (ablation) | SAS · Scenario I (large S(beta), x_new^L) | rMSE: 0.269 -> 0.296 | Hurt |
| make true risk model dense (larger S(beta)) (ablation) | supervised LASSO · Scenario I AUC | AUC: 0.694 -> 0.660 | Hurt |
| make true risk model dense (larger S(beta)) (ablation) | SAS · Scenario I AUC | AUC: 0.713 -> 0.711 | No clear effect |
| use SAS debiased inference instead of debiased SLASSO | 95% CI for sparse x_new^S · Scenario II (approx. sparse, moderate surrogates) | coverage probability: 0.784 -> 0.944 | Improved |
| use SAS debiased inference instead of debiased SLASSO | 95% CI for dense x_new^D · Scenario I (approx. sparse model) | coverage probability: 0.812 -> 0.944 | Improved |
| use strong instead of moderate surrogates | debiased SAS · Scenario I (x_new^L) | empirical standard error: 1.873 -> 1.531 | Improved |
Closest papers by meaning
- Efficient Modeling of Surrogates to Improve Multi-source High-dimensional Integrative Regression (JMLR 2026 · similarity 0.90)
- Prior Adaptive Semi-supervised Learning with Application to EHR Phenotyping (JMLR 2022 · similarity 0.90)
- Democratizing Clinical Risk Prediction with Cross-Cohort Cross-Modal Knowledge Transfer (NeurIPS 2025 · similarity 0.86)
- SMART: Towards Pre-trained Missing-Aware Model for Patient Health Status Prediction (NeurIPS 2024 · similarity 0.83)
- High-dimensional Analysis of Knowledge Distillation: Weak-to-Strong Generalization and Scaling Laws (ICLR 2025 · similarity 0.83)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.82)
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.
