Collaborative Research on Kidney Exchange

当有人希望将肾脏捐赠给特定的接受者时，肾脏交换的需求就会出现。当一对供体与另一对供体相容时（或在涉及更多不相容的供体的循环中），两对或更多这样不相容的供体有时可以通过交换接受移植。部分源于NSF先前资助的研究，有组织的肾脏交换，如新英格兰肾脏交换计划（neke）已经在美国变得活跃起来，现在多个地区的肾脏交换已经存在，并且正在增长，一个全国性的肾脏交换正在考虑中（随着2007年Charlie W.Norwood活体器官捐赠法案的通过）。包括许多医院在内的肾脏交换的增长提出了新的实践和理论问题，该奖项资助研究开发机制设计的新方法来回答这些问题。第一个问题是激励机制。当肾脏交换项目开始时，第一批模型包括患者-供体配对和他们的外科医生。然而，区域交流的成功表明，移植中心（医院）也必须纳入战略分析。移植中心处理多对患者-供体配对，他们的策略包括可能不让一些配对进行肾脏交换，只透露那些医院自己无法匹配的配对。该团队在实践中观察到了这一点，初步的研究结果表明，在使用有效的肾脏交换匹配算法的情况下，不可能让医院透露所有的配对。进一步的研究探讨了大群体中匹配机制的激励特性。pi试图证明，即使在大量人口中，现有机制也不具有良好的激励特性，但当池中有许多患者-捐赠者对时，可以设计出既具有良好激励又具有良好效率特性的替代机制，就像成功的国家交换的情况一样。项目的第二部分涉及效率。pi将Erdos和r<s:1>尼（1959,1966）的经典（无向）随机图结果改编为模拟患者-供体对之间兼容性的图，并使用这些图来探索具有许多对的大型肾脏交换的行为。这项工作的第一部分将关注患者和供体的一般人群，第二部分将考虑当有大量极度敏感的患者找到供体的可能性很低时产生的不同结构的大图表。（这将涉及对兼容性图进行更详细的建模；不再是为每个患者和她的供体提供一个节点，而是必须为患者和供体提供单独的节点，因为供体的连通性将不受患者的影响。）找到合适的肾脏的可能性很低，因此，一旦找到这样的肾脏，这对配对就会很好地联系在一起。我们还将考虑不同的肾脏交换策略，包括管理可能较长的捐赠链。更广泛的影响：在美国有超过80,000名患者在等待尸体肾脏的名单上。2009年，33,678名患者加入了等待名单，10,441例尸体肾脏移植手术进行了，4,456名患者在等待名单上死亡，超过1,941名患者被从名单上删除。病得不能移植？在美国，还有6387例活体肾脏移植。可供移植的肾脏严重短缺。肾脏交换是近年来仍然很小但增长最快的活体肾脏供体来源。改进用于匹配供体和受者的方法将挽救生命。pi将与移植中心和区域联盟合作，将新开发的方法付诸应用。

The need for kidney exchange arises when someone who wishes to donate a kidney to a particular recipient is incompatible with that recipient. Two or more such incompatible patient-donor pairs can sometimes both receive transplants through an exchange,when the donor in one pair is compatible with the recipient in the other (or in cycles involving more incompatible pairs). Stemming in part from previously funded NSF research, organized kidney exchanges, such as the New England Program for Kidney Exchange (NEPKE) havebecome active in the United States, and now multiple regional kidney exchanges exist and are growing, and a national exchange is being contemplated (following the passage of the Charlie W.Norwood Living Organ Donation Act in 2007). The growth of kidney exchange to include many hospitals has raised new practical and theoretical questions and this award funds research that develops new methods in mechanism design to answer these questions.The first issue is incentives. When kidney exchange programs started, the first models incorporatedpatient-donor pairs and their surgeons. However, the success of regional exchanges has demonstrated that transplant centers (hospitals) must also be included in the strategic analysis. Transplant centers deal with multiple patient-donor pairs, and their strategy sets include the possibility of withholding some pairs from the kidney exchange and revealing only those pairs that the hospital cannot match on its own. The team has observed exactly this in practice, and preliminary research results suggest that there is no way to get hospitals to reveal all their pairs while still using a efficient kidney exchange matching algorithm. Additional research explores the incentive properties of matching mechanisms in large populations. The PIs seek to demonstrate that existing mechanisms do not have good incentive properties even in large populations, but that alternative mechanisms can be designed that would have both good incentives and good efficiency properties when there are many patient-donor pairs in the pool, as would be the case in a successful national exchange.The second part of the project concerns efficiency. The PIs adapt the classical (undirected) random graph results of Erdos and Rényi (1959, 1966) to graphs that model the compatibility between patient-donor pairs, and use these to explore the behavior of large kidney exchanges, with many pairs. The first part of this work will look at general populations of patients and donors, and the second part will consider the differently structured large graphs that result when there is a large population of extremely sensitized patients who have low probability of finding a donor. (This will involve modeling the compatibility graph in more detail; instead of one node for each patient and her donors, there will have to be separate nodes for patients and donors, since the connectedness of donors will be unaffected by the patient?s low probability offinding a compatible kidney, so that, once such a kidney is found, the pair will be well connected.) We will also consider different kidney exchange strategies, including the management of possibly long chains of donations.Broader Impacts: There are over 80,000 patients on the waiting list for cadaver kidneys in the U.S. In 2009, 33,678 patients were added to the waiting list,10,441 transplants of cadaver kidneys were performed, 4,456 patients died while on the waiting list, and more than 1,941 others were removed from the list as ?Too Sick to Transplant?. There were also 6,387 transplants of kidneys from living donors in the US. There is a severe shortage of kidneys for transplant. Kidney exchange is in recent years the still small but fastest growing source of live kidney donors. Improvements in the methods used to match donors and recipients will save lives. The PIs will collaborate with transplant centers and regional alliances to take the newly developed methods into application.