The transplant system already has one clear example of a formula that disadvantaged patients for years. It is the best guide to what can go wrong when AI tools enter the same system, and to who is supposed to catch it.
The Kidney Formula Doctors estimate kidney function with an equation called eGFR. For years, the equation raised the result for Black patients based on race, which made their kidney function look better than it was and delayed the point at which kidney failure was diagnosed and a transplant referral followed. In 2021 a National Kidney Foundation and American Society of Nephrology task force recommended removing race from the calculation.
The Fix In July 2022 UNOS prohibited race-based eGFR calculations at transplant hospitals, and in January 2023 it required programs to review their waitlists and adjust waiting time for affected Black patients. Pooja Singh of Thomas Jefferson University said her team adjusted the accrued wait time for 405 patients and that 122 of them received transplants sooner. No public tally of patients who died while waiting because of the old correction was found in the sources reviewed.
Why It Matters for AI The formula was a simple equation, not machine learning, and it still took years to change. A machine-learning model can be harder to inspect, because it learns patterns from past data, and past data reflects past decisions. The Microsoft Research team that studied kidney offer acceptance wrote that its best models did not clearly or equitably outperform existing allocation policies. Toni Eyssallenne of the Coalition to End Racism in Clinical Algorithms has said that changing an algorithm can take a decade or more.
3×3 Vinyl Sticker / Sliding scale $3.00–$8.00
The Oversight Gap GAO’s January 2026 report found that HRSA did not complete required annual performance evaluations of UNOS for years, began retroactive evaluations in March 2025, and had not assessed the risks of UNOS’s optional supplementary services, which brought in about $9.6 million in fiscal year 2024. GAO also said the coordination plan between HRSA and CMS lacked specific steps, milestone dates and success measures. HHS agreed with GAO’s three recommendations.
New Rules for Procurement Organizations CMS issued a final rule on Oct. 30, 2026 revising the conditions OPOs must meet to participate in Medicare and Medicaid. The tier system still grades OPOs on donation and transplantation rates, and an OPO with only bottom-tier service areas is decertified. The rule requires OPO quality programs to use data to monitor and improve performance, and it defines unsound medical practices that threaten patient safety as grounds for urgent decertification. The fact sheet names no specific technology requirements. In September 2025, HHS moved to decertify the Life Alliance Organ Recovery Agency in Miami after an investigation found unsafe practices and understaffing, and it announced a tool showing when organs are allocated outside the standard match list.
Nevada’s Stake Nevada Donor Network, the Las Vegas-based OPO, reported 795 organ transplants and 262 organ donors across the state in 2025, and CMS selected it in January 2026 to serve Southern Florida after the Miami agency was decertified. That makes how it records decisions and uses tools a question with reach beyond Nevada. The sources reviewed do not describe any machine-learning use by the organization.
Anyone considering registering as a donor should know that donor registration and organ allocation are separate steps, and that the safeguards discussed here apply after a donation, not to the choice to register. Watch for the HRSA rulemaking on enforceable OPTN policies and for any announcement that adds AI to the matching system, because that is where independent audits would matter most.
Discover more from KVIG Informative
Subscribe to get the latest posts sent to your email.
