Italiano (Italian) English (Inglese)
Friday, 23 June 2017


Pubblication Details
Authors:Marco Beccuti
Giuliana Franceschinis
Silvia Villa
Scientific Area:Performance Evaluation
Title:A dynamic simulation model for comparing kidney exchange policies
Published on:TR-INF-2017-03-01-UNIPMN
Publisher:DiSIT, Computer Science Institute, UPO
Tipo Pubblicazione:Technical Report
Abstract:In this paper we tackle the dynamic kidney exchange problem. First, we propose a formal definition through a mathematical model whose solution provides an optimal exchange policy, then we analyze two almost myopic kidney exchange policies which are currently adopted in practice by many transplant organizations. With respect to previous works in this area the models proposed in this paper take into account the quality of the performed exchanges. The dynamic kidney exchange problem is formulated through a Markov Decision Process (MDP) model which allows to evaluate the performance of an optimal policy in a simplified case (constraining the state space size of the MDP). The results obtained through numerical solution of the MDP are then compared with the performance of two almost myopic policies evaluated using simulation. The obtained results show that in this case the performance of the considered policies is quite similar to that of the optimal policy. The analysis is then extended to more realistic settings by means of a parametric simulator (freely available for download) that allows to consider multiple characteristics of both donors and recipients, different degrees of compatibility quantifying the quality of the exchange, and different cost factors concurring to the definition of the reward function on which the evaluation is based. The influence of the various parameters on the achievable reward is thoroughly discussed.