Research map / Label-efficient and robust learning / Recognition and long-tail
Diagnosing failures of fairness transfer across distribution shift in real-world medical settings
Jessica Schrouff, Natalie Harris, Oluwasanmi O Koyejo, Ibrahim Alabdulmohsin, Eva Schnider, Krista Opsahl-Ong, Alexander Brown, Subhrajit Roy, Diana Mincu, Chrsitina Chen, Awa Dieng, Yuan Liu and 5 more
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 |
|---|---|---|---|
| Deploy the dermatology model trained on source data in the Colombia target environment | ResNet dermatology model · Dermatology, source to Colombia target | Age-group top-3 accuracy gap: 1.05% to 12.86% | Hurt |
| Deploy the EHR risk model across generalist and specialized ICUs | EHR risk model · MIMIC-III, source to target ICUs | Demographic parity gap (sex): 0.002 ± 0.002 to 0.016 ± 0.003 | Hurt |
| Apply age-based demographic parity post-processing on the source model | EHR risk model · MIMIC-III target | Demographic parity gap (age): 10^{-5} to 0.078 ± 0.025 | Hurt |
| Apply age-based equalized odds post-processing on the source model | EHR risk model · MIMIC-III target | Equalized odds gap (age): 0.05 ± 0.008 to 0.16 ± 0.04 | Hurt |
| Use conditional-independence tests with parent-balancing weights to diagnose direct shift effects | Shift diagnosis procedure · Dermatology and MIMIC-III | Detection of shift-affected variables | Improved |
| Use source-only fairness mitigation under a compound shift | Fairness mitigation methods · MIMIC-III target | Fairness transfer: age gap increases to 11.4% | Hurt |
Closest papers by meaning
- Rethinking Fair Representation Learning for Performance-Sensitive Tasks (ICLR 2025 · similarity 0.86)
- FairPFN: A Tabular Foundation Model for Causal Fairness (ICML 2025 · similarity 0.86)
- Fairness under Covariate Shift: Improving Fairness-Accuracy Tradeoff with Few Unlabeled Test Samples (AAAI 2024 · similarity 0.86)
- Understanding challenges to the interpretation of disaggregated evaluations of algorithmic fairness (NeurIPS 2025 · similarity 0.85)
- The Boundaries of Fair AI in Medical Image Prognosis: A Causal Perspective (NeurIPS 2025 · similarity 0.85)
- The Fragility of Fairness: Causal Sensitivity Analysis for Fair Machine Learning (NeurIPS 2024 · 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.
