Sequential Decision Processes in Kidney Paired Donation Programs with Altruistic Donors

• Activity Number: 666
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Yijiang Li*+ and John Kalbfleisch and Peter Song and Alan Leichtman
• Companies: University of Michigan and University of Michigan and University of Michigan and University of Michigan
• Keywords: altruistic donors ; Decision under uncertainty ; integer programming ; look-ahead search tree ; probabilistic modeling

Abstract

In this talk we focus on a kidney paired donation (KPD) program comprised of incompatible donor/candidate pairs and living non-directed (or altruistic) donors. An altruistic donor donates to the candidate of an incompatible pair with the understanding that the donor associated with the benefitted-candidate will donate to the candidate of another pair in the program, and so on. The central task is to determine an organ allocation strategy under uncertainties that yields an optimal (or nearly optimal) expected value of a pre-specified objective function. Analogous to the way a computer plays chess, we propose a look-ahead search tree to sequentially allocate the altruistic donor so as to maximize the expected utility over a certain given number of moves. Our proposed strategy is evaluated over a series of match runs in simulations where donors and candidates are generated based on data from actual KPD systems. Objective functions used in simulations include the expected overall number of transplants achieved, as well as utilities that incorporate a candidate's waiting time for a transplant and expected outcomes (e.g. survival and graft survival) associated with a possible transplan