Research map / Label-efficient and robust learning / Domain adaptation and generalization
Exploring Accurate and Transparent Domain Adaptation in Predictive Healthcare via Concept-Grounded Orthogonal Inference
Pengfei Hu, Chang Lu, Feifan Liu, Yue Ning
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 SAE reconstruction and domain classification losses with orthogonal inference (ablation) | MMD-aligned encoder · eICU diagnosis | weighted F1: 64.87 -> 68.61 | Improved |
| add orthogonal inference (W, z) and domain classification loss (ablation) | ExtraCare · eICU diagnosis | weighted F1: 66.73 -> 68.61 | Improved |
| use dictionary-induced metric M instead of Euclidean inner product (ablation) | ExtraCare · eICU diagnosis | weighted F1: 67.58 -> 68.61 | Improved |
| supervise covariate residual with domain classification loss rather than adversarial loss (ablation) | ExtraCare · OCHIN diagnosis | weighted F1: 71.28 -> 74.05 | Improved |
| add M-metric orthogonal residual on OCHIN heart failure (ablation) | ExtraCare · OCHIN heart failure | AUROC: 95.07 -> 95.48 | No clear effect |
| ExtraCare vs best baseline on eICU heart failure F1 | BUA · eICU heart failure | F1: 79.83 -> 79.64 | No clear effect |
| apply ExtraCare relative to source-only model | Base · OCHIN diagnosis | weighted F1: 63.77 -> 74.05 | Improved |
| use simple DANN or DAL alignment | Base · eICU diagnosis | weighted F1: 61.09 -> 62.14 (DANN), 61.58 (DAL) | No clear effect |
| use self-training (CST, SSRT) for adaptation | Base · eICU and OCHIN | various | Mixed |
| use ExtraCare sparse code explanations instead of regular SAE | regular SAE · OCHIN facility adaptation | Label-Axis Difference (%): 36 -> 47 | Improved |
| use ExtraCare explanations instead of regular SAE | regular SAE · OCHIN facility adaptation | Domain-Axis Difference (%): 40 -> 42 | No clear effect |
| use ExtraCare explanations instead of regular SAE | regular SAE · OCHIN facility adaptation | Quadrant Calibration Error (%), lower better: 37 -> 32 | Improved |
Closest papers by meaning
- One Loss to Rule Them All: Marked Time-to-Event for Structured EHR Foundation Models (NeurIPS 2026 · similarity 0.85)
- Risk Horizons: Structured Hypothesis Spaces for Longitudinal Clinical Prediction (NeurIPS 2026 · similarity 0.85)
- A Textbook Remedy for Domain Shifts: Knowledge Priors for Medical Image Analysis (NeurIPS 2024 · similarity 0.84)
- DrFuse: Learning Disentangled Representation for Clinical Multi-Modal Fusion with Missing Modality and Modal Inconsistency (AAAI 2024 · similarity 0.84)
- Context Clues: Evaluating Long Context Models for Clinical Prediction Tasks on EHR Data (ICLR 2025 · similarity 0.84)
- BoxLM: Unifying Structures and Semantics of Medical Concepts for Diagnosis Prediction in Healthcare (ICML 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.
