32nd Annual Student Research Conference
Quality and Fairness of Online Matching Algorithms for Kidney Exchange
Kelsey Lieberman
Dr. Jon Beck, Dr. Sanmay Das (Washington University in St. Louis), and Dr. Chien-Ju Ho (Washington University in St. Louis), Faculty Mentors
Kidneys are unique amongst organs in that a transplant can come from either a living or a deceased donor. One option for patients in need of a kidney with a willing living, but incompatible, donor is to enter an exchange pool. Recent research has shown that adding compatible pairs to the pool drastically improves incompatible pairs’ chances of receiving a match in 2 or 3 cycles. We design a new algorithm, Online Dual Assignment using Shadow Survival Estimates (ODASSE) to optimize online matching in the market for the total expected graft survival (EGS) using a data simulator based on parameters from a major transplant center. The algorithm performs linear regression on demographic information and dual variables predicted on the pool to predict shadow survival estimates. We find that this algorithm consistently outperforms greedy algorithms and algorithms optimizing for count. Additionally, we show that this algorithm improves outcomes for harder-to-match pairs.
Keywords: Kidney exchange, Fairness, Online matching
Topic(s):Computer Science
Biology
Economics
Presentation Type: Oral Paper
Session: 103-5
Location: BH 212
Time: 9:30