Aligning with Heterogeneous Preferences for Kidney Exchange

• URL: http://arxiv.org/abs/2006.09519v1
• Date: Tue, 16 Jun 2020 21:16:53 GMT
• Title: Aligning with Heterogeneous Preferences for Kidney Exchange
• Authors: Rachel Freedman
• Abstract summary: We propose a methodology for prioritizing patients based on heterogeneous moral preferences. We find that this methodology increases the average rank of matched patients in the sampled preference ordering, indicating better satisfaction of group preferences.
• Score: 7.858296711223292
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: AI algorithms increasingly make decisions that impact entire groups of humans. Since humans tend to hold varying and even conflicting preferences, AI algorithms responsible for making decisions on behalf of such groups encounter the problem of preference aggregation: combining inconsistent and sometimes contradictory individual preferences into a representative aggregate. In this paper, we address this problem in a real-world public health context: kidney exchange. The algorithms that allocate kidneys from living donors to patients needing transplants in kidney exchange matching markets should prioritize patients in a way that aligns with the values of the community they serve, but allocation preferences vary widely across individuals. In this paper, we propose, implement and evaluate a methodology for prioritizing patients based on such heterogeneous moral preferences. Instead of selecting a single static set of patient weights, we learn a distribution over preference functions based on human subject responses to allocation dilemmas, then sample from this distribution to dynamically determine patient weights during matching. We find that this methodology increases the average rank of matched patients in the sampled preference ordering, indicating better satisfaction of group preferences. We hope that this work will suggest a roadmap for future automated moral decision making on behalf of heterogeneous groups.

Related papers

• ComPO: Community Preferences for Language Model Personalization [122.54846260663922]
  ComPO is a method to personalize preference optimization in language models. We collect and release ComPRed, a question answering dataset with community-level preferences from Reddit.
  arXiv  Detail & Related papers  (2024-10-21T14:02:40Z)
• Towards Stable Preferences for Stakeholder-aligned Machine Learning [0.48533995158972176]
  The primary objective of this study is to create a method for learning both individual and group-level preferences pertaining to kidney allocations. By incorporating stakeholder preferences into the kidney allocation process, we aspire to advance the ethical dimensions of organ transplantation.
  arXiv  Detail & Related papers  (2024-01-27T02:21:31Z)
• Machine learning-based patient selection in an emergency department [0.0]
  This paper investigates the potential of an Machine Learning (ML) based patient selection method. It incorporates a comprehensive state representation of the system and a complex non-linear selection function. Results show that the proposed method significantly outperforms the APQ method for a majority of evaluated settings.
  arXiv  Detail & Related papers  (2022-06-08T08:56:52Z)
• Finding Regions of Heterogeneity in Decision-Making via Expected Conditional Covariance [3.9775905909091804]
  We present an algorithm for identifying types of contexts with high inter-decision-maker disagreement. We formalize this as a causal inference problem, seeking a region where the assignment of decision-maker has a large causal effect on the decision. We apply our algorithm to real-world healthcare datasets, recovering variation that aligns with existing clinical knowledge.
  arXiv  Detail & Related papers  (2021-10-27T15:20:12Z)
• A Twin Neural Model for Uplift [59.38563723706796]
  Uplift is a particular case of conditional treatment effect modeling. We propose a new loss function defined by leveraging a connection with the Bayesian interpretation of the relative risk. We show our proposed method is competitive with the state-of-the-art in simulation setting and on real data from large scale randomized experiments.
  arXiv  Detail & Related papers  (2021-05-11T16:02:39Z)
• Learning $\mathbf{\mathit{Matching}}$ Representations for Individualized Organ Transplantation Allocation [98.43063331640538]
  We formulate the problem of learning data-driven rules for organ matching using observational data for organ allocations and transplant outcomes. We propose a model based on representation learning to predict donor-recipient compatibility. Our model outperforms state-of-art allocation methods and policies executed by human experts.
  arXiv  Detail & Related papers  (2021-01-28T01:33:21Z)
• Adversarial Sample Enhanced Domain Adaptation: A Case Study on Predictive Modeling with Electronic Health Records [57.75125067744978]
  We propose a data augmentation method to facilitate domain adaptation. adversarially generated samples are used during domain adaptation. Results confirm the effectiveness of our method and the generality on different tasks.
  arXiv  Detail & Related papers  (2021-01-13T03:20:20Z)
• Indecision Modeling [50.00689136829134]
  It is important that AI systems act in ways which align with human values. People are often indecisive, and especially so when their decision has moral implications.
  arXiv  Detail & Related papers  (2020-12-15T18:32:37Z)
• Adapting a Kidney Exchange Algorithm to Align with Human Values [59.395925461012126]
  We provide an end-to-end methodology for estimating weights of individual participant profiles in a kidney exchange. We show how to use these weights in kidney exchange market clearing algorithms.
  arXiv  Detail & Related papers  (2020-05-19T21:00:29Z)
• Online Batch Decision-Making with High-Dimensional Covariates [20.06690325969748]
  We propose and investigate a class of new algorithms for sequential decision making that interact with textita batch of users simultaneously instead of textita user at each decision epoch. We deliver a solution, named textitTeamwork LASSO Bandit algorithm, that resolves a batch version of explore-exploit dilemma via switching between stage and selfish stage during the whole decision process.
  arXiv  Detail & Related papers  (2020-02-21T17:36:15Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.