Pollack, Madeleine, Daillak, Thomas, Toner, Mehmet, Uygun, Korkut, Yeh, Heidi, Gupta, Swati · Oct 28, 2026
Pollack and coauthors propose an AI framework for kidney-allocation decisions
Markdown · JSON · Original source
Madeleine Pollack and coauthors ask when AI trained on past decisions should recommend an action, defer to human experts, or flag patterns that warrant deviating from past practice. Their case study is deceased-donor kidney allocation, where out-of-sequence (OOS) allocation can help place kidneys at risk of discard but bypasses candidates higher on the waitlist. The authors argue OOS allocation should be used only when a kidney is demonstrably at high risk of discard through the standard sequence. Their Arbitrator framework combines interpretable models and acts only when uncertainty in predictions cannot change the preferred action; otherwise, it defers. On match-run data from 2016–2019 and 2023–2025, the authors report that it resolved 93.5% of placement decisions and deferred 6.5%. They also report identifying 70.1% of historically discarded low-KDPI kidneys as OOS candidates, projecting that up to 43.9% of historically discarded kidneys could be flagged for expedited placement and potentially transplanted. Those figures describe the study’s analysis and projections, not demonstrated outcomes in clinical practice. The authors also use the framework to compare learned rules with published clinical literature and flag contentious patterns. The page says e-prints are not peer-reviewed and should not guide clinical practice without context.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
cs.AI · Article
Source details
Original source date: Oct 28, 2026 (2026-10-28).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-11T11:38:07.857Z. First observed: 2026-10-30T06:42:47.128Z. Retrieved: 2026-10-31T08:48:33.355Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The evidence is an abstract, not the full paper; the page warns that arXiv e-prints are not peer-reviewed and should not guide clinical practice without context. The reported percentages and projection come from match-run data and historical decisions; the supplied evidence does not establish outcomes from clinical deployment.
Open question
What evidence would establish that a kidney is at high risk of discard through the standard sequence, the condition the authors set for OOS allocation?
Source-stated
Pollack and coauthors propose a framework for deciding when AI trained on historical decisions should recommend, defer, or flag patterns that depart from past practice.
We ask when an AI trained on historical decisions should be allowed to recommend decisions, defer to human experts, or in other cases, flag concerning patterns in data and deviate from past practice.
Source-stated
The authors say OOS allocation should be reserved for kidneys demonstrably at high risk of discard on the standard sequence, because it bypasses candidates higher on the waitlist.
OOS allocation should only be used when the organ is demonstrably at high risk of discard on the standard sequence.
Source-stated
The Arbitrator combines interpretable models and acts only when uncertainty cannot change the preferred action; otherwise it defers.
we build the Arbitrator framework, which combines various interpretable models and acts only when the uncertainty in their predictions cannot change the preferred action.
Source-stated
On the study data, the authors report that the Arbitrator resolved 93.5% of placement decisions and deferred 6.5%.
Our Arbitrator resolves 93.5% of organ placement decisions using explicit medical criteria learned from data, and defers the remaining 6.5% to a human or a more complex model.
Source-stated
The authors project that up to 43.9% of historically discarded kidneys could be flagged for expedited placement and potentially transplanted; this is a projection, not a reported transplant outcome.
projecting that up to 43.9% of historically discarded kidneys could be flagged for expedited placement and potentially transplanted.
Source and contributor context
Pollack, Madeleine, Daillak, Thomas, Toner, Mehmet, Uygun, Korkut, Yeh, Heidi, Gupta, Swati · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: 27fa1e38-baab-4299-821d-0db56c6024db.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-11T11:38:07.857Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: d29596f942f6494ef77a7e773c7f0f14ed03655ee7330b07daec18cfaf80a817. Source text hash: 05402eddf9e18644720a2516bb9586d8c1a082a9925c9115a36c29e1aa9a774a.
Revision and source-check history
Version 1 · 2026-11-11T11:38:07.857Z
Pollack and coauthors propose an AI framework for kidney-allocation decisions
Madeleine Pollack and coauthors ask when AI trained on past decisions should recommend an action, defer to human experts, or flag patterns that warrant deviating from past practice. Their case study is deceased-donor kidney allocation, where out-of-sequence (OOS) allocation can help place kidneys at risk of discard but bypasses candidates higher on the waitlist. The authors argue OOS allocation should be used only when a kidney is demonstrably at high risk of discard through the standard sequence. Their Arbitrator framework combines interpretable models and acts only when uncertainty in predictions cannot change the preferred action; otherwise, it defers. On match-run data from 2016–2019 and 2023–2025, the authors report that it resolved 93.5% of placement decisions and deferred 6.5%. They also report identifying 70.1% of historically discarded low-KDPI kidneys as OOS candidates, projecting that up to 43.9% of historically discarded kidneys could be flagged for expedited placement and potentially transplanted. Those figures describe the study’s analysis and projections, not demonstrated outcomes in clinical practice. The authors also use the framework to compare learned rules with published clinical literature and flag contentious patterns. The page says e-prints are not peer-reviewed and should not guide clinical practice without context.
Original source
cs.AI · Article
Source details
Original source date: Oct 28, 2026 (2026-10-28).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-11T11:38:07.857Z. First observed: 2026-10-30T06:42:47.128Z. Retrieved: 2026-10-31T08:48:33.355Z.
Limits
The evidence is an abstract, not the full paper; the page warns that arXiv e-prints are not peer-reviewed and should not guide clinical practice without context. The reported percentages and projection come from match-run data and historical decisions; the supplied evidence does not establish outcomes from clinical deployment.
Open question
What evidence would establish that a kidney is at high risk of discard through the standard sequence, the condition the authors set for OOS allocation?
Source-stated
Pollack and coauthors propose a framework for deciding when AI trained on historical decisions should recommend, defer, or flag patterns that depart from past practice.
We ask when an AI trained on historical decisions should be allowed to recommend decisions, defer to human experts, or in other cases, flag concerning patterns in data and deviate from past practice.
Source-stated
The authors say OOS allocation should be reserved for kidneys demonstrably at high risk of discard on the standard sequence, because it bypasses candidates higher on the waitlist.
OOS allocation should only be used when the organ is demonstrably at high risk of discard on the standard sequence.
Source-stated
The Arbitrator combines interpretable models and acts only when uncertainty cannot change the preferred action; otherwise it defers.
we build the Arbitrator framework, which combines various interpretable models and acts only when the uncertainty in their predictions cannot change the preferred action.
Source-stated
On the study data, the authors report that the Arbitrator resolved 93.5% of placement decisions and deferred 6.5%.
Our Arbitrator resolves 93.5% of organ placement decisions using explicit medical criteria learned from data, and defers the remaining 6.5% to a human or a more complex model.
Source-stated
The authors project that up to 43.9% of historically discarded kidneys could be flagged for expedited placement and potentially transplanted; this is a projection, not a reported transplant outcome.
projecting that up to 43.9% of historically discarded kidneys could be flagged for expedited placement and potentially transplanted.
Source and contributor context
Pollack, Madeleine, Daillak, Thomas, Toner, Mehmet, Uygun, Korkut, Yeh, Heidi, Gupta, Swati · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: 27fa1e38-baab-4299-821d-0db56c6024db.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-11T11:38:07.857Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: d29596f942f6494ef77a7e773c7f0f14ed03655ee7330b07daec18cfaf80a817. Source text hash: 05402eddf9e18644720a2516bb9586d8c1a082a9925c9115a36c29e1aa9a774a.
uncertain · 2026-11-11T11:38:07.857Z · anthropic/claude-opus-5.5
Publication: 27fa1e38-baab-4299-821d-0db56c6024db
attribution — pass: The draft correctly credits Madeleine Pollack and coauthors. It keeps the framework, the OOS condition and all reported figures attributed to the authors as their reported analysis and projections.
Authors:Madeleine Pollack, Thomas Daillak, Mehmet Toner, Korkut Uygun, Heidi Yeh, Swati Gupta
summary — uncertain: The draft is faithful but omits the source's central motivation: nearly 30% of recovered US deceased-donor kidneys are never transplanted. Without it, a general reader cannot see why the problem matters. It also leaves 'low-KDPI' and 'match-run data' unexplained, and drops that the data is US OPTN data excluding COVID-disrupted years. Smallest repair: add one sentence on the roughly 30% non-transplant figure with US/OPTN scope. Add tooltips for KDPI and match-run.
Nearly 30% of kidneys recovered from deceased donors in the United States are never transplanted, largely because they are hard to place through the standard waitlist sequence.
date — pass: originalPublished is 2026-10-28, which matches the v1 submission date on the arXiv page. The later feed date is correctly not used.
[Submitted on 28 Oct 2026]
support — pass: Each claim is directly supported by the source text. The draft labels 43.9% as a projection, not a demonstrated outcome, and notes the absence of peer review and clinical deployment evidence. The question follows from the authors' own stated condition for OOS allocation without presupposing an answer.
Thus, OOS allocation should only be used when the organ is demonstrably at high risk of discard on the standard sequence.
media — pass: The selected media is the verified direct arXiv abstract page for this paper. It is the most direct access to the source.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXiv