Public Notices

The organ allocation process: a natural extension of the Carrel Agent-Mediated Electronic Institution

Description
The organ allocation process: a natural extension of the Carrel Agent-Mediated Electronic Institution
Categories
Published
of 12
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  1 The organ allocation process: a naturalextension of the Carrel Agent MediatedElectronic Institution Javier V´azquez-Salceda    , Ulises Cort´es    ,Julian Padget ✁  , Antonio L´opez-Navidad ✂  ,Francisco Caballero ✂  , ✄ Technical University of Catalonia.c/Jordi Girona 1 & 3. 08034 Barcelona, Spain. E-mail: ☎  jvazquez,ia ✆  @lsi.upc.es ✝  Department of Computer Science. University of  Bath. Bath, BA2 7AY. United Kingdom. E-mail: jap@cs.bath.ac.uk  ✞  Banc de Teixits.Hospital de la Santa Creu i Sant Pau. c/ St. Antoni M. Claret, 167. E08025 Barcelona,Spain. In this paper we extend the formalization of   Carrel , a virtualorganization for the procurement of tissues for transplant, todeal also with organs. We focus on the organ allocation pro-cess to show how it can be formalized with the ISLANDERspecification language and wealso present the firstversion of a mechanism to federate several geographically-distributedCarrel platforms with the objective of addressing some of thetechnical issues in the establishment of a pan-European tis-sue and organ distribution service. 1. Introduction Organ transplants are among the most complicatedmedical procedures performed today. Organ transplan-tation from human donors are becoming common life-saving therapies and the only option available whenthere is major damage to or a malfunction in an organ.Today, most donated organs and tissues come from pa-tients who are pronounced brain dead as result of dis-ease or injury. At the time of writing, more than onemillion people in the world have successfully receivedan organ, and thereafter, in most cases, been able tolead normal lives. But while these operations are be-coming more commonplace, there is a major shortageof organs. At the time of writing, ten people die dailydue to the shortage of transplantable organs, while aOrgan Allowable time from Donorto RecipientHeart 4 to 5 (hours)Lung 5 to 6 (hours)Heart-Lung 4 to 6 (hours)Pancreas 12 to 15 (hours)Liver 12 to 18 (hours)Kidney 15 to 18 (hours) Fig. 1. Cold ischaemia time for solid organ transplantation new name is added to the transplant waiting list every18 minutes.Over the years, transplant techniques have evolved,knowledge of donor-recipient compatibility has im-provedand so haveimmuno-suppressant drug regimes,leading not only to an increase in the different kinds of organs that can be transplanted, but also in the range of transplants, moving beyond organs (heart, liver, lungs,kidney, pancreas) to tissues (bones, skin, corneas, ten-dons). The allocation process for tissues is quite dif-ferent from that for organs, because of the time organscan be preserved outside the human body (see table 1).Tissues are clusters of relativelyhomogeneous cells,so the optimal temperature for preservation of all thecells composing the tissue is almost the same. Thus,tissues can be preserved for several days (from sevendays in the case of corneas to years in the case of bones) in tissue banks. For tissues, the allocation pro-cess is demand-driven,triggered when there is a recip-ient with a need for a certain tissue, at which time sev-eral tissue banks are searched for a suitable piece.Organs, on the other hand, are very complex struc-tures with several kinds of cell types each with differ-ent optimal preservation temperatures. That fact leadsto quite short preservation times (hours), no need foran organbank, and an allocation process that is supply-driven, triggered when a donor appears, taking theform of a search for a suitable recipient in some num-ber of hospitals. AI CommunicationsISSN 0921-7126, IOS Press. All rights reserved  2 V´azquez-Salceda et al. / The organ allocation process: a natural extension of the Carrel Agent Mediated Electronic Institution 1.1. The need of software systems for the organ and tissue management  As explained in [22], the increasing rate of successof tissue transplants, that is literally a  second chance at life , is leading to an increase in the number of requestsandthisvolumeis startingtooverwhelmthehumanco-ordinators at hospitals who are responsible for manag-ing the transplant process, and furthermore is leadingto tissue loss, because available tissues are not beingassigned before they exceed their  shelf life  as a resultof the length of time it is taking to process requests.In the case of organs, successful transplants havealso led to an increase in demand for organs for trans-plantation purposes. However, there is not an increas-ing volume of donations to match the demand and thisis creatinglistsof transplantcandidatesor  waitinglists .Typically, there is one list per kind of organ. Becauseof the organshortage, the process of managing and dis-tributing the organs that are available is complex andoften surrounded by controversy. Much research hasbeen done to enable the definition and implementa-tion of policies for donor identification (to increase thenumber of available donors) [10], organ allocation (tofind a suitable recipient for each organ) [20,7,16] andin extraction, preservation and implant procedures (toincrease the chances of success).The relative scarcity of donors has led to the cre-ation of international coalitions of transplant organiza-tions. This new, more geographically distributed, envi-ronment makes an even stronger case for the applica-tion of distributed software systems to solve: –  the data exchangeproblem:  exchange of informa-tion is a major issue, as each of the actors col-lects different information and stores it in differ-ent formats. The obvious, and easily stated, solu-tion is the definition of standard data interchangeformats. –  the communication problem:  countries typicallyuse different languages and terminologies to tagthe same items or facts. Either a standard notationor a standard ontology or even a translation mech-anism is needed to avoid misunderstandings. –  the coordination issues:  in order to manage re-quests at an international level, there is the needto coordinate geographically distributed surgeryteams, and to coordinate piece delivery at an in-ternational level. –  the variety of regulations:  an additional issue isthe necessity of accommodating a complex set of,in some cases conflicting, national and interna-tional regulations, legislation and protocols gov-erning the exchange of organs [20,7]. These regu-lations also change over time, making it essentialthat the software is adaptable.The first two points can largely be resolved by stan-dard software solutions. For instance, the EU projects RETRANSPLANT  and  TECN  have largely focused onthe creation of a) standard formats for the storage andexchange of information about pieces, donors and re-cipients among organizations, b) telematic networks,or c) distributed databases . Another project  ESCU-LAPE  uses conventional software to help in checkingtissue histocompatibility.In the USA the United Network for Organ Sharing(UNOS) is supporting tools like ULAM [19] that areused for the simulation and analysis of national cadav-eric kidneyand kidney-pancreas allocation policies fortransplantation to permit comparison of multiple liverallocation policy proposals so that the policies can betested prior to implementation.The third point (coordination) is harder to solvewithconventional software. But a sound, if relatively new,approach is the use of   software agents . An  Agent   is acomputer program capable of taking its own decisionswith no external control ( autonomy ), based on its per-ceptions of the environment and the objectives it aimsto reach [23]. It not only reacts to the environment ( re-activity ) but also  pro-actively  takes initiatives. The  so-cial ability  of agents allow them to group together (in agencies ) sharing common objectives and dividing thetasksinordertoachievethoseobjectives.Alltheseuse-fulattributessuggest thatmulti-agent systemsare well-suited for solving coordination issues.It is the last point (variety of regulations changingover time) which underpins our case for the use of so-called  electronic institutions , whose purpose is to pro-vide over-arching frameworks for interaction of agentscapable of reasoning about the norms governing indi-viduals’ actions, in the same way as physical institu-tions and social norms do in the real world (see ✟  2).Electronic institutions and the norms that govern themare the key to a system that is able to adapt automati-cally to changes in regulations.In summary, our proposal addresses all four issuesidentified above, by the use of multi-agent technology,not only for coordination and regulation but also forserving as a language interface among teams using dif-ferent terminology, and actively distributing the infor-mation to be shared.  V´azquez-Salceda et al. / The organ allocation process: a natural extension of the Carrel Agent Mediated Electronic Institution 3 1.2. Organization of the paper  In ✠  2 we introduce our point of view of electronicinstitutions as  social structures  and the use of norms.In ✠  3 we give a description of   Carrel  our electronic or-ganization for the procurement of human organs andtissues for transplantation purposes. Here, we focus onthe organ allocation process to show how it can beformalized with the ISLANDER [4] specification lan-guage. In ✠  4 we discuss the main characteristics of  Carrel  and finally present some conclusions in ✠  5. 2. Electronic Institutions Building a Multi-Agent System (MAS) is a complextask, as developers need to find an acceptable balancebetween the  autonomy  that agents should have in or-der to act in unforeseen circumstances, and the  control the designer wants to have over emergent behaviour inorder to ensure that the system will achieve its goals.One way to tame the complexity of building a MASis to create a centralized controller, that is, a specificagent that ensures coordination.  Coordinator agents are agents which have some kind of control over otheragents’ goals or, at least on part of the work assignedto an agent, according to the knowledge about the ca-pabilities of each agent that is under the  Coordinator  Agent  ’s command. However this approach is not good,as this special agent becomes a  bottleneck   in the in-formation flow. An alternative is to distribute not onlythe work load but also the control among all the agentsin the system ( distributed control ). That means to  in-ternalize  control of each agent, which has now to beprovided with reasoning and social abilities to make itable to reason about intentions and knowledge of otheragents plustheglobalgoals of thesociety inorder to beable to coordinate successfully with others and also toresolve conflicts when they arise. However, as Mosesand Tennenholtz state in [13], in those domains wherethe cost of a conflict is dear, or if conflict resolution isdifficult, completely independent behaviour becomesunreasonable. Therefore some kind of structure shouldbe defined in order to ease coordination in a  distributed control  scenario. A good option taken from animal in-teractions is the definition of   social structures . Social structures 1 , define a social level where themulti-agent system is seen as a  society  of entities in 1 They are also called  Artificial Social Systems  by Shoham, Mosesand Tennenholtz in [13,18] order to enhance the coordination of agent activities(such as message passing management and the alloca-tion of tasks and resources) by defining structured pat-terns of behaviour. Social structures reduce the dangerof combinatorial explosion in dealing with the prob-lems of agent cognition, cooperation and control, asthey impose restrictions to the agents’ actions. Theserestrictions have a positive effect, as they: –  avoid potential conflicts, or ease their resolution –  makeeasierforagivenagenttoforeseeandmodelthe other agents’ behaviour in a closed environ-ment and fit its own behaviour accordingly.An  Agent-Mediated Electronic Institution  ( e-institu-tion  for short) is kind of   social structure  where the in-teractions among a group of agents are governed by aset of explicit norms expressed in a language represen-tation that agents can interpret.The roots of this idea come from the study of humanorganizations. Most human interactions are governedby conventions or rules of some sort, having their ori-gin in society (emergent) or the laws (codification of emergent rules) that society has developed. Thus wefind that all human societies, even the most primitiveones, have some kind of   social  constraints upon theirmembers in order to structure and regulate the rela-tions among their members. Some of these constraintsare quite informal (taboos, customs, traditions) whilesome others are formally defined (written laws, consti-tutions). In fact, modern human societies have definedcollections of expected behavioural patterns that havean effect in specific scenarios (such as a shop, bank, aconversation, a lecture or an exclusive club).Douglas North 2 refers to this  corpora  of constraintsas  institutions . In his studies [15], North has analyzedthe effect of institutions on the behaviour of humanorganizations (including human societies) and he con-cludes that institutional constraints ease human inter-action (reducing the cost of this interaction by ensuringtrust), shaping choices and making outcomes foresee-able. By the creation of these constraints, the organiza-tions and the interactions theyrequire can each growincomplexity while interaction costs remain static or areeven reduced. Having established these institutionalconstraints, every competent participant in the institu-tion may act—and expect others to act—according toa list of rights, duties, and protocols of interaction.The main reason grounding the creation of institu-tions is to create  trust   when the parties know very lit- 2 Douglas North received the Nobel Prize in 1993 for his studieson the role of institutions in the performance of organizations.  4 V´azquez-Salceda et al. / The organ allocation process: a natural extension of the Carrel Agent Mediated Electronic Institution tle about others. No institutions are necessary in anenvironment where parties have complete informationabout others ( e.g.  a village market where vendors andbuyers knoweach other and interact on a periodical ba-sis). However, in environments with incomplete infor-mation ( e.g.  international commerce), cooperative so-lutions (based on trust) could break down unless insti-tutions are created to provide sufficient information forindividuals to create trust and to police deviations.The same statements and ideas can be brought tothe field of agents. An  e-Institution  is the modellingof an institution through the specification of its normsin some suitable formalism(s) that can be followed byagents. One of the hypotheses of the field is that theessence of an institution, through its norms and pro-tocols can be captured in a precise machine process-able form (this key idea forms the core of the nascenttopic of institutional modelling). The main objective isto create a  safe  environment where agents can trust inother agents, as any violation of a norms may lead to acompensating sanction.The effect of norms on agents should not be seen asa constraint but a guide which reduces the complexityof theenvironment,becausetheoption space is pruned,and hence allows the agent to make better use of its(limited) resources. 3. An Institution for the distribution of organs andtissues The Carrel agent platform models an  organization that receives a tissue request from one hospital andthen tries to allocate the  best   tissue available from allthe known tissue banks. Within this organization dif-ferent entities play different roles that are determinedby their goals at any particular moment. Figure 2 de-picts all the parties that interact with the Carrel system.Therearea)thehospitals thatcreatethetissuerequests,b) the tissue banks, and c) the national organ trans-plantation organizations, that own the agent platformand act as observers (in the figure the organizationsin Spain are depicted: the Organizaci´on Nacional deTransplantes 3 (ONT) [17] and the Organitzaci`o CATa-lana de Transplantaments 4 (OCATT)). In the proposedsystem all hospitals, even those running a tissue bank,must make their requests through Carrel in order to en-sure a fair distribution of pieces and to ease the track-ingof allpiecesfromextractiontothetransplant, as theONT and OCATT require for organs(that is, no  insider  trading). 3 National Transplant Organization 4 Catalan Transplant OrganizationFig. 2. Carrel: An Agent Mediated Institution for Tissues Assign-ment The role of the Carrel institution can be summarizedin the following tasks: –  to make sure that all the agents which enter intothe institution behave properly (that is, that theyfollow the behavioural norms). –  to be up to date about all the available pieces inthetissue banks,andall therecipients thatare reg-istered in the waiting lists. –  to check that all hospitals and tissue banks fulfilthe requirements for interacting with Carrel. –  to ensure that the commitments made within Car-rel are fulfilled. –  to coordinate the piece delivery from one facilityto another. –  to record all incidents relating to a particularpiece.The participation of hospitals in Carrel is based onthe notion of membership. That is, hospitals belong toCarrel and respect the negotiation (assignation) rules,and the agents that represent them inside Carrel are un-able to break these conventions. A Hospital interactswith Carrel through the Transplant Coordination UnitAgency ( UCTx ). This agency (depicted in figure 16)serves as interface between the surgeons and Carrel.When a surgeon needs a piece, he makes his requestthrough the UCTx system, which analyzes the infor-mation entered, adds information about the recipientand finally creates a  Finder Agent  , which goes to theinstitution looking for a suitable piece.The information required by the  Finder Agent   tolook for a piece in  Carrel  is entered in an electronic Sealed Envelope . The envelope contains the followinginformation: –  Urgency level , that works as an electronic postagestamp and sets the urgencylevel of the request (inSpain: normal, urgency-1 or urgency-0)  V´azquez-Salceda et al. / The organ allocation process: a natural extension of the Carrel Agent Mediated Electronic Institution 5 –  Hospital identification , a certificate issued by theCertification Authority associated with Carrel [2],to allow the authentication of the sender of eachrequest to ensure that only  Finder Agents  with re-quests from authorizedhospitals can enter and ne-gotiate. –  Tissue information  (type, parameters,  etc. ) and  re-cipient data  (age, sex, laboratory analysis,  etc. ). –  The  selection function  which is composed of aset of rules, each one being a constraint the se-lected piece ( e.g.  a cornea) must satisfy. Some of these rules will typically come from the policy of the transplant unit of the hospital, while the restwill be introduced by the surgeon, who can set theconstraints needed for a given recipient. A rule of the  selection function  can contain: ✡ predicates about the piece: predicates thatdescribe the constraints the selected tissuehas to satisfy, such as the age of the donoror the dimensions of the piece itself. ✡ predicates about the tissue bank: predicatesthat can set constraints about the tissue bank preferred by the surgeon or the hospital. ✡ predicates about the cost of the piece: apredicate that can set a maximum cost forthepiece.Thiscostonlycoverspiece extrac-tion and preservation, and is paid througha clearing house by the hospital which re-ceivesthepiece. Thereis no price associatedwith the piece itself. 3.1. Extending the Carrel institution In order to extend the Carrel system presented in[22] to handle organ as well as tissue distribution, wemust augment it for the organ allocation process. Inmost of the official organ allocation organizations, theprocess is composed of two phases:1. Each hospital informs the organization about pa-tients that have been added to or removed fromthe waiting list of that hospital, or patients eitherto be added to or removed from the national-wideMaximum Urgency Level 5 Waiting List.2. When a donor appears, the hospital informs theorganization of all the organs suitable for dona-tion in the form of   offers  sent to the organ alloca-tion organization, which then assigns the organs. 5 In Spain the Maximum Urgency Level is called Urgency-0 This process can be modelled formally by specify-ing interactions between agents. To give a formal de-scription of the interaction among agents in the Car-rel system we will follow the same formalism usedfor the case of tissues [22]. The ISLANDER formal-ism [4] views an agent-based electronic institution asa type of   dialogical system  where all the interactionsinside the institution are the composition of multipledialogic activities (message exchanges). These inter-actions (called  illocutions  [14]) are structured throughagent group meetings called  scenes  that follow well-defined protocols.However, instead of creating a separate model forthe organ allocation process, we will extend the modelfor the tissue allocation process. Some of the scenesthat were defined for the case of tissues will be sharedfor organs by extending their functionalities, and a fewnew scenes are created. The resulting set of scenes isthe following: –  Reception Room : the scene where all the exter-nal agents identify themselves in order to be as-signed the roles they are authorized to play. If these agents are carrying either a request for oneor more tissues or an offer of one or more organs,then this information is checked to make sure thatit is well-formed. –  Consultation Room : the scene where the institu-tion is updated about any event or incident relatedto a piece. Agents coming from tissue banks up-date the institution about tissue availability, whileagents coming from hospitals update the institu-tion about waiting lists and information on piece(organ or tissue) reception, transplant operationsand the health of recipients. –  Exchange Room : the scene where the assignationprocess is made. In fact, there are specific ex-change rooms for managing tissue requests ( Tis-sue Exchange Room ) and for organ offers ( Organ Exchange Room ). –  Confirmation Room : the scene where the provi-sional assignments made in either a  Tissue Ex-change Room  or a  Organ Exchange Room  areconfirmed or cancelled because of the arrival of another request with higher priority.In case of theconfirmed ones, a delivery plan is built.Another key element of the ISLANDER formalismis the definition of agent  roles . Each agent can be asso-ciatedwithoneor moreroles,andtheserolesdefinethescenes the agent can enter and the protocols it shouldfollow (the  scene protocols  are defined as multi-roleconversational patterns). There are two kinds of roles:
Search
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x