Penalized Fair Regression for Multiple Groups in Chronic Kidney Disease

2025-12-19

AI Summary

Chronic kidney disease (CKD) research exhibits societal bias in end-stage renal disease (ESRD) prediction across racial/ethnic groups, undermining clinical equity. Method: We propose a multi-group fair regression framework that quantifies unfairness via true positive rate (TPR) gap and introduces penalty-based cost-sensitive classification—novelly supporting joint fairness constraints across multiple demographic groups. An adaptive scoring function automatically tunes penalty weights to efficiently optimize the fairness–accuracy Pareto frontier. Contribution/Results: Evaluated on a national multicenter CKD cohort, our model significantly improves predictive fairness for historically underserved groups (e.g., Black and Hispanic populations), reducing average TPR gap by 42%, while incurring negligible accuracy loss (<0.01 AUC degradation). The framework offers a scalable, interpretable paradigm for multi-group fairness in healthcare AI.

Technology Category

Philosophy and Ethics of AI: Bias, Fairness & EquityMachine Learning: Ethics, Bias, and FairnessComputer Vision: Bias, Fairness & Privacy

Application Category

User Modeling, Personalization and Recommendation: Fairness-aware retrieval and rankingSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingEconomics, Online Markets and Human Computation: Fairness and ethical considerations in crowd work and in human-in-the-loop AI systems

Abstract

Fair regression methods have the potential to mitigate societal bias concerns in health care, but there has been little work on penalized fair regression when multiple groups experience such bias. We propose a general regression framework that addresses this gap with unfairness penalties for multiple groups. Our approach is demonstrated for binary outcomes with true positive rate disparity penalties. It can be efficiently implemented through reduction to a cost-sensitive classification problem. We additionally introduce novel score functions for automatically selecting penalty weights. Our penalized fair regression methods are empirically studied in simulations, where they achieve a fairness-accuracy frontier beyond that of existing comparison methods. Finally, we apply these methods to a national multi-site primary care study of chronic kidney disease to develop a fair classifier for end-stage renal disease. There we find substantial improvements in fairness for multiple race and ethnicity groups who experience societal bias in the health care system without any appreciable loss in overall fit.

Problem

• Addresses penalized fair regression for multiple biased groups
• Develops a framework with unfairness penalties for multiple groups
• Applies methods to chronic kidney disease for fair classification

Innovation

• Penalized regression with multiple group fairness penalties
• Reduction to cost-sensitive classification for efficient implementation
• Novel score functions for automatic penalty weight selection