Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Multi-Attribute Fairness in Healthcare Predictive Modeling

Xiaoyang Wang, Christopher C. Yang

TL;DR

The work tackles multi-attribute fairness in healthcare predictive modeling by first optimizing predictive performance and then applying fairness-driven fine-tuning across multiple demographic attributes via sequential and simultaneous strategies. Using a transfer-learning framework, the approach demonstrates notable reductions in Equalized Odds Disparity across Race and Sex while preserving competitive AUROC on two real-world healthcare datasets (SUD and sepsis mortality). The study analyzes trade-offs between sequential and simultaneous fairness optimization, showing that sequential prioritizes the first attribute but simultaneous yields more balanced fairness with similar predictive performance. These findings advance equitable healthcare AI by highlighting the importance of addressing multiple demographic dimensions concurrently rather than in isolation, and they point to practical directions for extending fairness methods to more complex, multi-modal, and multi-class settings.

Abstract

Artificial intelligence (AI) systems in healthcare have demonstrated remarkable potential to improve patient outcomes. However, if not designed with fairness in mind, they also carry the risks of perpetuating or exacerbating existing health disparities. Although numerous fairness-enhancing techniques have been proposed, most focus on a single sensitive attribute and neglect the broader impact that optimizing fairness for one attribute may have on the fairness of other sensitive attributes. In this work, we introduce a novel approach to multi-attribute fairness optimization in healthcare AI, tackling fairness concerns across multiple demographic attributes concurrently. Our method follows a two-phase approach: initially optimizing for predictive performance, followed by fine-tuning to achieve fairness across multiple sensitive attributes. We develop our proposed method using two strategies, sequential and simultaneous. Our results show a significant reduction in Equalized Odds Disparity (EOD) for multiple attributes, while maintaining high predictive accuracy. Notably, we demonstrate that single-attribute fairness methods can inadvertently increase disparities in non-targeted attributes whereas simultaneous multi-attribute optimization achieves more balanced fairness improvements across all attributes. These findings highlight the importance of comprehensive fairness strategies in healthcare AI and offer promising directions for future research in this critical area.

Enhancing Multi-Attribute Fairness in Healthcare Predictive Modeling

TL;DR

The work tackles multi-attribute fairness in healthcare predictive modeling by first optimizing predictive performance and then applying fairness-driven fine-tuning across multiple demographic attributes via sequential and simultaneous strategies. Using a transfer-learning framework, the approach demonstrates notable reductions in Equalized Odds Disparity across Race and Sex while preserving competitive AUROC on two real-world healthcare datasets (SUD and sepsis mortality). The study analyzes trade-offs between sequential and simultaneous fairness optimization, showing that sequential prioritizes the first attribute but simultaneous yields more balanced fairness with similar predictive performance. These findings advance equitable healthcare AI by highlighting the importance of addressing multiple demographic dimensions concurrently rather than in isolation, and they point to practical directions for extending fairness methods to more complex, multi-modal, and multi-class settings.

Abstract

Artificial intelligence (AI) systems in healthcare have demonstrated remarkable potential to improve patient outcomes. However, if not designed with fairness in mind, they also carry the risks of perpetuating or exacerbating existing health disparities. Although numerous fairness-enhancing techniques have been proposed, most focus on a single sensitive attribute and neglect the broader impact that optimizing fairness for one attribute may have on the fairness of other sensitive attributes. In this work, we introduce a novel approach to multi-attribute fairness optimization in healthcare AI, tackling fairness concerns across multiple demographic attributes concurrently. Our method follows a two-phase approach: initially optimizing for predictive performance, followed by fine-tuning to achieve fairness across multiple sensitive attributes. We develop our proposed method using two strategies, sequential and simultaneous. Our results show a significant reduction in Equalized Odds Disparity (EOD) for multiple attributes, while maintaining high predictive accuracy. Notably, we demonstrate that single-attribute fairness methods can inadvertently increase disparities in non-targeted attributes whereas simultaneous multi-attribute optimization achieves more balanced fairness improvements across all attributes. These findings highlight the importance of comprehensive fairness strategies in healthcare AI and offer promising directions for future research in this critical area.
Paper Structure (31 sections, 16 equations, 1 figure, 7 tables, 2 algorithms)

This paper contains 31 sections, 16 equations, 1 figure, 7 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Preliminary
  3. Related Work
  4. Group Fairness in Machine Learning
  5. Demographic Parity
  6. Equalized Odds and Equal Opportunity
  7. Calibration
  8. Bias Mitigation
  9. Fairness in Healthcare AI
  10. Methods and Materials
  11. Performance Optimization
  12. Fairness Optimization
  13. Sequential Strategy
  14. Simultaneous Strategy
  15. Experiments
  16. ...and 16 more sections

Figures (1)

  • Figure 1: The multi-attribute fairness optimization pipeline, illustrating the performance optimization phase followed by the fairness optimization phase using sequential and simultaneous strategies.