Causal Fairness Framework for Survival Analysis
A new research paper introduces a causal framework for fairness in survival analysis, a temporal machine learning task that predicts time-to-event outcomes. The work addresses a gap in fair ML, which has largely focused on static settings. Current fair survival methods rely on statistical definitions that cannot disentangle causal mechanisms behind disparities. The proposed framework decomposes survival disparities into direct, indirect, and other causal contributions. The paper is available on arXiv under ID 2605.11362.
Key facts
- Paper title: Causal Fairness for Survival Analysis
- arXiv ID: 2605.11362
- Announce type: cross
- Focuses on fairness in survival/time-to-event analysis
- Proposes a causal framework to decompose disparities
- Addresses gap in fair ML for temporal contexts
- Critiques statistical fairness definitions as insufficient
- Published on arXiv
Entities
Institutions
- arXiv