Adapting a Kidney Exchange Algorithm to Align With Human Values
Authors: Rachel Freedman, Jana Schaich Borg, Walter Sinnott-Armstrong, John Dickerson, Vincent Conitzer
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We then evaluate the impact of the weights in simulations and find that the precise numerical values of the weights we computed matter little, other than the ordering of profiles that they imply. |
| Researcher Affiliation | Academia | Rachel Freedman Duke University rachel.freedman@duke.edu Jana Schaich Borg Duke University js524@duke.edu Walter Sinnott-Armstrong Duke University ws66@duke.edu John P. Dickerson University of Maryland john@cs.umd.edu Vincent Conitzer Duke University conitzer@cs.duke.edu |
| Pseudocode | No | The paper describes the algorithm mathematically (integer program formulation) and textually, but does not provide a structured pseudocode block or an explicitly labeled algorithm. |
| Open Source Code | Yes | All code for this paper can be found in the Ethics package of github.com/John Dickerson/Kidney Exchange. |
| Open Datasets | No | Based on previously developed tools (Dickerson and Sandholm 2015), we built a simulator to mimic daily matching in a real-world kidney exchange pool. The demographics of our simulated pool were designed to reflect the UNOS kidney exchange pool where possible, and otherwise the general US population. They also collected their own human subjects data from MTurk but do not provide access details for this collected data. |
| Dataset Splits | No | The paper describes running simulations over a period of 5 simulated years, but it does not specify traditional train/validation/test dataset splits. It also collects human subject data but doesn't mention splits for that data. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU/GPU models, memory specifications) used for running the experiments or simulations. |
| Software Dependencies | Yes | We used the BTm() function in the Bradley Terry2 package in R to estimate profile scores p1, . . . , p8 based on the 8092 pairwise comparisons, both directly and as a function of the estimated scores of their three attribute values. |
| Experiment Setup | Yes | We ran 20 simulations of daily matching over the course of 5 simulated years using both algorithms. The paper also details the attributes, their alternatives (Table 1), and the estimated profile scores (Table 3), which are key parameters for their algorithm setup. |