Fairness-Aware Kidney Exchange and Kidney Paired Donation
Mingrui Zhang, Xiaowu Dai, Lexin Li · 2025
Read Fairness-Aware Kidney Exchange and Kidney Paired Donation by Mingrui Zhang, Xiaowu Dai, Lexin Li (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-06431.txt.
Introduction
Mingrui Zhang, Xiaowu Dai, and Lexin Li add a new fairness rule to kidney exchange matching, in a 2025 paper. Paired donation swaps kidneys between incompatible donor-patient pairs. Existing fairness ideas cover groups and individuals. They ignore legally protected traits like race and gender. The authors borrow a calibration idea from machine learning. Match chances should not depend on protected traits once medical sensitivity is fixed.
Their rule enters the matching optimizer as a constraint. Linearization and column generation keep solving practical. Theory prices the fairness cost on random graph models. Simulations and a real data case compare the three fairness styles. The new rule protects traits the older rules overlook. Costs appear modest in their tests. The framework speaks to allocation systems beyond kidneys as well.
Catalog record
| CREATOR | Mingrui Zhang, Xiaowu Dai, Lexin Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 92 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-06431
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Fairness-Aware Kidney Exchange and Kidney Paired Donation Mingrui Zhang, Xiaowu Dai, Lexin Li Abstract The kidney paired donation (KPD) program provides an innovative solution to overcome incompatibility challenges in kidney transplants by matching incompatible donor-patient pairs and facilitating kidney exchanges. To address unequal access to transplant opportunities, there are two widely used fairness criteria: group fairness and individual fairness. However, these criteria do not consider protected patient features, which refer to characteristics legally or ethically recognized as needing protection from discrimination, such as race and gender. Motivated by the calibration principle in machine learning, we introduce a new fairness criterion: the matching outcome should be conditionally independent of the protected feature, given the sensitization level. We integrate this fairness criterion as a constraint within the KPD optimization framework and propose a computationally efficient solution using linearization strategies and column-generation methods. Theoretically, we analyze the associated price of fairness using random graph models. Empirically, we compare our fairness criterion with group fairness and individual fairness through both simulations and a real-data example. Keywords : calibration, column generation, integer program, kidney paired donation, price of fairness, random graph. Introduction 1.1 Kidney paired donation programs Kidney transplantation is the preferred treatment for end-stage renal disease (ESRD), offering significant improvements in both quality of life and survival compared to dialysis. However, a major obstacle is the incompatibility between donors and patients, often due to mismatches in blood type or human leukocyte antigens (HLA). According to the United Network for Organ Sharing (UNOS) and the Organ Procurement and Transplantation Network (OPTN), over 90,000 patients were on the kidney transplant waiting list at the end of 2023. To overcome these incompatibility challenges, kidney paired donation (KPD) programs have been developed as an innovative solution. These programs match incompatible donor-patient pairs using a virtual crossmatch , a preliminary compatibility test, and facilitate kidney exchanges, allowing patients to receive kidneys from compatible donors through mutual exchange. These exchanges are called exchange cycles or simply cycles , with formal definitions provided in Section 2 . The primary goal of KPD programs is to maximize the number of successful transplants or optimize generalized utilities based on predicted transplantation or survival outcomes. Despite their promise, planned cycles may fail for various reasons, such as illness, pregnancy, or the death of a patient or donor, scheduling conflicts, or discrepancies between virtual and laboratory crossmatch results. These uncertainties necessitate the consideration of recourse strategies, which identify alternative transplant opportunities within the original cycle. Even when an entire cycle cannot proceed, smaller unaffected sub-cycles may still be viable. By accounting for these uncertainties, KPD programs can further maximize the expected number of successful transplants or optimize the expectation of some general utilities. Following the framework of Klimentova et al. (2016) , KPD programs adopt three main recourse strategies. The first is no recourse , where a cycle either proceeds fully or fails entirely ( Li et al. 2014 , e.g.) . The second, internal recourse , identifies the sub-cycle with the highest utility among those unaffected by the failure ( Pedroso 2014 , e.g.) . The third, subset recourse , considers broader subsets that may include multiple cycles, enabling alternative arrangements when uncertainties arise ( Bray et al. 2018 , e.g.) . Each strategy computes expected utilities accordingly, with KPD programs aiming to maximize these expected utilities. In addition to recourse strategies, we can integrate uncertainties into the optimization framework through other approaches, such as look-ahead strategy ( Wang et al. 2017 ) and robust optimization ( McElfresh et al. 2019 ) . 1.2 Fairness concerns in KPD programs Despite the success of KPD programs, important fairness concerns arise, particularly regarding unequal access to transplant opportunities. These disparities stem from two main factors. The first factor is differences in patients’ HLA sensitization levels. A patient’s HLA sensitization level is measured by their panel-reactive antibody (PRA) score, which reflects the likelihood of HLA incompatibility with a random donor. Patients with high PRA scores, referred to as highly sensitized patients, are more difficult to match and, therefore, have fewer transplant opportunities. In contrast, patients with low PRA scores, known as lowly-sensitized patients, are easier to match and benefit from more transplant opportunities. The second factor is asymmetric blood type compatibility. According to standard ABO blood type compatibility rules, patients with blood type O are harder to match because they can only receive kidneys from O-type donors. Conversely, patients with blood type AB are the easiest to match, as they can receive kidneys from donors of any blood type. To address these disparities, KPD programs can incorporate fairness constraints to reduce unfairness caused by differences in HLA sensitization and blood type. Two widely used fairness criteria in KPD programs are group fairness and individual fairness. Group fairness focuses on ensuring that highly-sensitized patients receive equitable consideration relative to lowly-sensitized patients ( Dickerson et al. 2014 ; McElfresh et al. 2019 ; Freedman et al. 2020 ) . In contrast, individual fairness aims to provide balanced selection chances for each patient, ensuring that no one is unfairly disadvantaged ( Farnadi et al. 2021 ) . While these two fairness criteria are central to addressing disparities in kidney exchange, other approaches have also been explored. For example, St-Arnaud et al. (2022) incorporate the Nash standard of comparison (or proportional fairness) and Rawlsian justice principles. Ashlagi and Roth (2014) , Klimentova et al. (2021) , and Carvalho and Lodi (2023) draw on game theory to address fairness within utility-maximization frameworks, considering the interests of stakeholders such as hospitals and regions. In the context of fair machine learning, a protected feature (or sensitive attribute) refers to a characteristic legally or ethically recognized as needing protection from discrimination. Our key question is how to establish a fairness criterion that ensures equal access to transplant opportunities across patient groups defined by protected characteristics—and how to achieve this in practice. This specific focus has not yet been explored in the KPD literature, as existing fairness criteria do not account for protected patient features. Some protected features, such as race and gender, are associated with differences in sensitization levels and blood types. For instance, studies have shown that parous women are more likely to develop high sensitization to HLA antigens ( Bromberger et al. 2017 ) , making them less compatible with most donors in a KPD program and harder to match. This leads to unequal access to transplant opportunities between females and males. A simplistic approach might aim to balance overall selection rates between genders, but this could inadvertently disadvantage highly-sensitized male patients, as a higher number of highly-sensitized female patients would need to be matched to achieve gender balance. A more equitable approach would balance selection rates within subgroups, such as highly-sensitized females versus males and lowly-sensitized females versus males. Motivated by this example, we propose a new fairness criterion: the matching outcome should be conditionally independent of the protected feature, given the sensitization level. The randomness associated with this fairness criterion is determined by a randomization policy, as proposed for individual fairness in kidney exchange ( Farnadi et al. 2021 ; St-Arnaud et al. 2022 ) and general matching problems ( García-Soriano and Bonchi 2020 ; Karni et al. 2022 ) .