Learning Social Fairness Preferences from Non-Expert Stakeholder Opinions in Kidney Placement

TL;DR

This work tackles fairness in ML-driven kidney placement by incorporating non-expert (public) opinions to learn social fairness preferences. It introduces a logit-based fairness feedback model and a SAFF (Social Aggregation of Fairness Feedback) algorithm that minimizes social feedback regret to infer a population-wide fairness weight vector over six group fairness notions. The approach is validated with simulated data and a Prolific survey using the kidney acceptance rate predictor (ARP) applied to OPTN/STAR data, revealing that public fairness priorities lean toward accuracy-related notions (AE, PE) and that ARP is perceived as reasonably fair overall. The findings offer a path to align ML-based organ offer decision tools with diverse stakeholder values, potentially reducing discrimination in kidney placement while maintaining practical predictive performance.

Abstract

Modern kidney placement incorporates several intelligent recommendation systems which exhibit social discrimination due to biases inherited from training data. Although initial attempts were made in the literature to study algorithmic fairness in kidney placement, these methods replace true outcomes with surgeons' decisions due to the long delays involved in recording such outcomes reliably. However, the replacement of true outcomes with surgeons' decisions disregards expert stakeholders' biases as well as social opinions of other stakeholders who do not possess medical expertise. This paper alleviates the latter concern and designs a novel fairness feedback survey to evaluate an acceptance rate predictor (ARP) that predicts a kidney's acceptance rate in a given kidney-match pair. The survey is launched on Prolific, a crowdsourcing platform, and public opinions are collected from 85 anonymous crowd participants. A novel social fairness preference learning algorithm is proposed based on minimizing social feedback regret computed using a novel logit-based fairness feedback model. The proposed model and learning algorithm are both validated using simulation experiments as well as Prolific data. Public preferences towards group fairness notions in the context of kidney placement have been estimated and discussed in detail. The specific ARP tested in the Prolific survey has been deemed fair by the participants.

Figures (9)

• Figure 1: An Example of Recipient Characteristics
• Figure 2: Four Questions Presented to the Participants for Each Data Tuple
• Figure 3: Non-Expert Participant's Feedback Model
• Figure 4: Social Aggregation of Fairness Feedback
• Figure 5: SAFF