Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fairness Perceptions in Regression-based Predictive Models

Mukund Telukunta, Venkata Sriram Siddhardh Nadendla, Morgan Stuart, Casey Canfield

TL;DR

This paper addresses bias and fairness concerns in regression-based organ-offer analytics for kidney transplantation by introducing three KL-divergence-based, regression-appropriate group-fairness notions: Independence, Separation, and Sufficiency, defined via $\mathbb{D}_{KL}$. It then couples these notions with a fairness-feedback framework and a Mixed-Logit–based social-preference learner (SAFF) to estimate public fairness preferences and minimize social-feedback regret. Through a crowdsourced study (N=85) and simulations, the authors find strong public preference for Separation and Sufficiency, with UPAT perceived as fair for gender and race but not for age, and demonstrate convergence of the SAFF learning process within about 50 epochs. These insights provide a principled approach to auditing and aligning regression-based decision-support tools in healthcare, with implications for public trust and policy in kidney allocation.

Abstract

Regression-based predictive analytics used in modern kidney transplantation is known to inherit biases from training data. This leads to social discrimination and inefficient organ utilization, particularly in the context of a few social groups. Despite this concern, there is limited research on fairness in regression and its impact on organ utilization and placement. This paper introduces three novel divergence-based group fairness notions: (i) independence, (ii) separation, and (iii) sufficiency to assess the fairness of regression-based analytics tools. In addition, fairness preferences are investigated from crowd feedback, in order to identify a socially accepted group fairness criterion for evaluating these tools. A total of 85 participants were recruited from the Prolific crowdsourcing platform, and a Mixed-Logit discrete choice model was used to model fairness feedback and estimate social fairness preferences. The findings clearly depict a strong preference towards the separation and sufficiency fairness notions, and that the predictive analytics is deemed fair with respect to gender and race groups, but unfair in terms of age groups.

Fairness Perceptions in Regression-based Predictive Models

TL;DR

This paper addresses bias and fairness concerns in regression-based organ-offer analytics for kidney transplantation by introducing three KL-divergence-based, regression-appropriate group-fairness notions: Independence, Separation, and Sufficiency, defined via . It then couples these notions with a fairness-feedback framework and a Mixed-Logit–based social-preference learner (SAFF) to estimate public fairness preferences and minimize social-feedback regret. Through a crowdsourced study (N=85) and simulations, the authors find strong public preference for Separation and Sufficiency, with UPAT perceived as fair for gender and race but not for age, and demonstrate convergence of the SAFF learning process within about 50 epochs. These insights provide a principled approach to auditing and aligning regression-based decision-support tools in healthcare, with implications for public trust and policy in kidney allocation.

Abstract

Regression-based predictive analytics used in modern kidney transplantation is known to inherit biases from training data. This leads to social discrimination and inefficient organ utilization, particularly in the context of a few social groups. Despite this concern, there is limited research on fairness in regression and its impact on organ utilization and placement. This paper introduces three novel divergence-based group fairness notions: (i) independence, (ii) separation, and (iii) sufficiency to assess the fairness of regression-based analytics tools. In addition, fairness preferences are investigated from crowd feedback, in order to identify a socially accepted group fairness criterion for evaluating these tools. A total of 85 participants were recruited from the Prolific crowdsourcing platform, and a Mixed-Logit discrete choice model was used to model fairness feedback and estimate social fairness preferences. The findings clearly depict a strong preference towards the separation and sufficiency fairness notions, and that the predictive analytics is deemed fair with respect to gender and race groups, but unfair in terms of age groups.
Paper Structure (20 sections, 28 equations, 7 figures, 5 tables)

This paper contains 20 sections, 28 equations, 7 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Experiment Design
  3. Datasets and Preprocessing
  4. Participant Survey Design
  5. Participants Demographics
  6. Method
  7. Divergence-Based Group Fairness Notions
  8. Fairness Feedback Model
  9. Fairness Preference Learning Algorithms
  10. Experiment Setup
  11. Results
  12. Simulation Results
  13. Survey Results
  14. Discussion on Survey Results
  15. Limitations
  16. ...and 5 more sections

Figures (7)

  • Figure 1: An Example of Recipient Characteristics
  • Figure 2: Three Questions Presented to the Participants for Each Data Tuple
  • Figure 3: Overview of the Proposed Fairness Feedback Model
  • Figure 4: Social Aggregation of Fairness Feedback
  • Figure 5: Convergence of Feedback Regret in Simulation with $M = 10$ Tuples.
  • ...and 2 more figures

Theorems & Definitions (3)

  • Definition 1: Independence
  • Definition 2: Separation
  • Definition 3: Sufficiency