Human-centered design of a distributed knowledge management system

  • Published on
    04-Sep-2016

  • View
    214

  • Download
    1

Embed Size (px)

Transcript

<ul><li><p>fen</p><p>hy. S</p><p>Texa</p><p>, Hou</p><p>nter,</p><p>Keywords:Human-centered computing; Organizational memory; Knowledge management; Collaborative; Groupware; Information systems; Design;</p><p>healthcare, however, the culture is still to train people has been on the design of the interfaces between systemsand users, such as usability testing, and not the deeperstructures that are fundamental for the design of trulyhuman-centered systems [1]. We argue that human-cen-tered computing goes beyond the representations in an</p><p>Journal of Biomedical Informat</p><p>.</p><p>* Corresponding author.E-mail addresses: susan_rinkus@yahoo.com (S. Rinkus), Jiajie.</p><p>Zhang@uth.tmc.edu (J. Zhang).1532-0464/$ - see front matter 2004 Elsevier Inc. All rights reservedEthnographic; Project lifecycle</p><p>1. Introduction</p><p>A large number of healthcare information technologyprojects fail. Many of these failures are not due toawed technology, but rather due to the lack of system-atic considerations of human issues in the systemsrequirements and specications processes. In otherindustries such as aviation, nuclear power plants, auto-mobiles, and consumer software and electronics, hu-man-centered design is commonly practiced. In</p><p>to adapt to poorly designed technology, rather than todesign technology to t peoples characteristics. System-atically incorporating human-centered design is neces-sary for successful development of information systemsthat increase eciency, productivity, ease of use, learn-ing, user adoption, retention, satisfaction, and decreasedevelopment time, support and training costs, and med-ical errors. The study of humancomputer interaction(HCI) has made signicant contributions to the designof user-friendly systems. However, the primary focusReceived 13 October 2004Available online 15 December 2004</p><p>Abstract</p><p>Many healthcare technology projects fail due to the lack of consideration of human issues, such as workow, organizationalchange, and usability, during the design and implementation stages of a projects development process. Even when human issuesare considered, the consideration is typically on designing better user interfaces. We argue that human-centered computing goesbeyond a better user interface: it should include considerations of users, functions and tasks that are fundamental to human-centeredcomputing. From this perspective, we integrated a previously developed human-centered methodology with a Project Design Life-cycle, and we applied this integration in the design of a complex distributed knowledge management system for the Biomedical Engi-neer (BME) domain in the Mission Control Center at NASA Johnson Space Center. We analyzed this complex system, identied itsproblems, generated systems requirements, and provided specications of a replacement prototype for eective organizational mem-ory and knowledge management. We demonstrated the value provided by our human-centered approach and described the uniqueproperties, structures, and processes discovered using this methodology and how they contributed in the design of the prototype. 2004 Elsevier Inc. All rights reserved.Human-centered design omanagem</p><p>Susan Rinkusa, Muhammad Waljia, KatJames P. Turleya, Jack W</p><p>a School of Health Information Sciences, University of</p><p>7000 Fannin, Suite 600b NASA Johnson Space Cedoi:10.1016/j.jbi.2004.11.014a distributed knowledget system</p><p>A. Johnson-Throopa,b, Jane T. Malinb,mitha,b, Jiajie Zhanga,*</p><p>s School of Health Information Sciences at Houston,</p><p>ston, TX 77030, USA</p><p>Houston, TX 77058, USA</p><p>www.elsevier.com/locate/yjbin</p><p>ics 38 (2005) 417</p></li><li><p>interface, but also includes users, functions and tasksthat are fundamental to the processes at large.</p><p>In this paper, we describe the application of a human-centered design methodology called human centered dis-tributed information design (HCDID) [1] to generate thesystems requirements and specications for a web-basedknowledge management system for biomedical engi-neers (BMEs) in Mission Control Center at the NASAJohnson Space Center. The HCDID methodology wasdeveloped with the aim of providing systematic princi-ples, guidelines, and procedures for the design of com-plex, highly ecient distributed human-centeredinformation systems. In this paper, we will rst describethe HCDID methodology. We will then describe theProject Design Lifecycle and how it is integrated withthe HCDID methodology. In this section we will dem-onstrate how the HCDID methodology and the ProjectDesign Lifecycle are used in the development of a web-based knowledge management system for NASAsBMEs. In the discussion and conclusion sections, we willdiscuss the value of our human-centered methodology inthe design of distributed information systems.</p><p>computing at the levels of users, functions, tasks, andrepresentations. As shown in Fig. 1, the components onthe left are multiple levels of analyses for single-userhuman-centered design. The user analysis level contrib-utes to each of the levels of functional, task, and represen-tational analysis. The components on the right representthe additional analysis needed for designing distributedhuman-centered information systems. The componentat the bottom represents the products of functional,task, and representational analyses. For each level ofanalysis, theHCDIDmethodology allows the researchersto employ several, alternate specic methods [1].</p><p>HCDID is based upon the principles of distributedcognition The core unit of analysis is the functional sys-tem which is composed of human and articial agentsand their relations which are distributed across timeand space dimensions [29]. Distributed cognition helpsdetermine which features of the activities or artifacts arerelevant for the eciency of task performance andwhich are necessary for the activity to continue to per-form well. It can identify complex interdependencies be-tween human and articial agents that occur incollaborative work environments and thus give theresearchers a better understanding of why simple break-</p><p>S. Rinkus et al. / Journal of Biomedical Informatics 38 (2005) 417 52. Methodology</p><p>2.1. Human-centered distributed information design</p><p>The human-centered distributed information design(HCDID) methodology considers human-centeredFig. 1. The human centered distributed information design (HCDID) methodowns in communications and interactions betweenthem can have such serious and signicant consequences[10].</p><p>HCDID provides a framework that addresses thedistributed social, cultural, organizational interactions,and cognitive issues involved in designing informationdology. (Simplied from Zhang et al., 2002, with authors permission.)</p></li><li><p>medictechnologies within a complex distributed, collaborativeenvironment. The following describes the componentsof the HCDID methodology.</p><p>2.1.1. User analysis</p><p>User analysis is the process of identifying the charac-teristics of existing and potential users, such as expertiseand skills, knowledge bases, educational background,cognitive capacities and limitations, perceptual varia-tions, age related skills, time available for learning andtraining, frequency of system use, etc. [11,12]. For ahealth information system, dierent users such as install-ers, administrators, nurses, physicians, registration per-sonnel, laboratory technicians, billing sta, andpatients may use dierent components of the system.Dierent users may also have dierent levels of under-standing of the same component of the system, such asbeginners, novices, and experts. User analysis helps usto design information systems that have the right knowl-edge and information structure that match those of theusers.</p><p>2.1.2. Functional analysis</p><p>Functional analysis is the process of identifying criti-cal top-level domain structures, goals, and inherentproperties of the work domain that are largely indepen-dent of implementations. It is more abstract than taskand representational analyses because it does not in-volve details of task processes and representation de-tails. For a distributed system, functional analysis alsoidenties the artifacts as well as articial and humanagents of the system, their interrelations and constraints,and their essential roles. For a knowledge-rich domainsuch as medicine or aviation, functional analysis re-quires extensive domain knowledge and a deep under-standing of domain structures.</p><p>2.1.3. Task analysisTask analysis is the process of identifying the proce-</p><p>dures and actions to be carried out and the informationto be processed to achieve task goals. One importantfunction of task analysis is to ensure that only the nec-essary and sucient task features that match userscapacities and are required by the task will be includedin systems specications. Extra fancy features that donot match users capacities or are not required by thetask will only generate extra processing demands forthe user and thus make the system harder to use. Fora distributed cognitive system it is important to performa distributed task analysis that identies the interactionsamong human and articial agents. This perspectivemay also help identify how multiple users interact withthe same data. The theory of distributed representationsdeveloped by Zhang and Norman [4,5] can be used toanalyze the distribution patterns of information among</p><p>6 S. Rinkus et al. / Journal of Biohuman and articial agents [13]. Task analysis can iden-tify overlooked tasks, the relative importance of tasks,the overlapping of task information, the grouping offunctions, the relation to user analysis, and so on. Itcan also pinpoint the bottlenecks or choking point ofthe task where special design has to be considered.</p><p>2.1.4. Representational analysis</p><p>Representational analysis is based upon a robust phe-nomenon called representational eect [5,14]: dierentrepresentations of a common abstract structure or pro-cess can generate dramatically dierent representationaleciencies, task diculties, and behavioral outcomes. Itis the process of identifying an appropriate informationdisplay format for a given task performed by a specictype of users such that the interaction between usersand systems is in a direct interaction mode [5,1517].With direct interaction interfaces, users can directly,completely, and eciently engage in the primary tasksthey intend to perform, not the housekeeping interfacetasks that are barriers between users and systems [18,19].</p><p>The form of a representation can inuence and some-times determine what information can be easily per-ceived, what processes are activated, what can bederived from the representation. For a complex noveltask, some portion of the task space may never be ex-plored and some structures of the task may never be dis-covered without a change in representation.</p><p>The end products generated from the methodology ofHCDID are the contents for the systems requirementsand specications of human-centered distributed infor-mation systems. Examples of these contents includefunctional requirements, goal-subgoal relations, taskstructures and procedures, information ow dynamics,and task-specic, event-related, and context-sensitiveinformation displays.</p><p>2.2. Project design lifecycle</p><p>A key advantage of the HCDID methodology is itspractical application to a projects design lifecycle. ThisProject Design Lifecycle is inherently iterative in natureand the theory and methods help to rene the productsgenerated from each phase. For discussion purposes, wehave placed the various components of the Project De-sign Lifecycle into specic phases. We recognize thatthese components occur throughout each phase andmay not be limited to any specic phase. It is importantto note that evaluation is a key step throughout the en-tire Project Design Lifecycle. The following describeshow the HCDID methodology is incorporated into thevarious stages of the Project Design Lifecycle (seeFig. 2).</p><p>2.2.1. Phase 1: Data collection and analysisThis phase seeks to discover key aspects about</p><p>al Informatics 38 (2005) 417the problem domain, their users, functions, and tasks.</p></li><li><p>S. Rinkus et al. / Journal of BiomedicDuring this phase, data collection and the user, func-tional, and task analyses in the HCDID frameworkare typically carried out. The products of this phase isthe identication of issues which are then framed withinan organizational memory and knowledge context</p><p>2.2.2. Phase 2: Systems requirements</p><p>The identication of the issues provides the buildingblocks to help dene the systems requirements. Func-tional analysis is the primary analysis during Phase 2.The products of this phase, the systems requirements,will then be mapped to provide the specications inPhase 3.</p><p>2.2.3. Phase 3: Specications</p><p>In this next phase, after the systems requirements aremapped to the specications, mockups are created. Rep-</p><p>Fig. 2. Project design lifecycle: HCDID theory and methods arecentral to all phases of the iterative lifecycle.resentational analysis from the HCDID methodologyplays a key role in helping to generate design alterna-tives. Dierent representations of a common abstractstructure or process can generate dramatically dierentrepresentational eciencies, task diculties, and behav-ioral outcomes. In addition to representational analysis,task analysis is also performed to supplement represen-tational analysis. Evaluations are continued throughoutthis phase and the systems requirements and specica-tions are further rened and used for designing the pro-totype in Phase 4.</p><p>2.2.4. Phase 4: Prototype</p><p>In this phase, a working prototype is developed fromthe mockups developed in Phase 3. Representation anal-ysis is conducted to help generate the user interface andto guide which representation is suited for each task.Usability testing is also performed throughout thisphase. After the completion of Phase 4, the cyclereiterates.3. Case study: biomedical engineering domain at NASA</p><p>In this section, we will describe a case study where weapplied the HCDID methodology and the Project De-sign Lifecycle to design a human-centered knowledgemanagement system. We will rst describe the domainof the Biomedical Engineers and the critical issues thatare central to human-centered information systems.Then we will describe the methods, procedures, and re-sults of applying the HCDID and the Project DesignLifecycle in the BME domain.</p><p>The task domain for the current study is the Biomed-ical Engineer console at Mission Control Center, NASAJohnson Space Center, Houston, Texas. In this domain,the primary roles are: (1) the Console BMEs who areresponsible for providing the technical and operationalsupport for medical operations activities involving theastronauts on the International Space Station; (2) theBME Liaisons (BME-L) who are responsible for track-ing and working on issues, and helping to reduce theinterruptions to the Console BME; and (3) Flight Sur-geons (FS) who have the primary authority for thehealth and safety of astronauts.</p><p>Astronauts in a space station are to some extent likepatients in an intensive care unit (ICU): their living envi-ronment is o nominal; their health conditions are mon-itored and evaluated continuousl...</p></li></ul>

Recommended

View more >