IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 8, NO. 3, SEPTEMBER 2004 279
Integrating Context-Aware Public Displays Into aMobile Hospital Information System
Jesus Favela, Marcela Rodrguez, Alfredo Preciado, and Victor M. Gonzlez
AbstractHospitals are convenient settings for deployment ofubiquitous computing technology. Not only are they technology-rich environments, but their workers experience a high level of mo-bility resulting in information infrastructures with artifacts dis-tributed throughout the premises. Hospital information systems(HISs) that provide access to electronic patient records are a step inthe direction of providing accurate and timely information to hos-pital staff in support of adequate decision-making. This has moti-vated the introduction of mobile computing technology in hospitalsbased on designs which respond to their particular conditions anddemands. Among those conditions is the fact that worker mobilitydoes not exclude the need for having shared information artifactsat particular locations. In this paper, we extend a handheld-basedmobile HIS with ubiquitous computing technology and describehow public displays are integrated with handheld and the servicesoffered by these devices. Public displays become aware of the pres-ence of physicians and nurses in their vicinity and adapt to provideusers with personalized, relevant information. An agent-based ar-chitecture allows the integration of proactive components that offerinformation relevant to the case at hand, either from medical guide-lines or previous similar cases.
Index TermsAutonomous agents, context-aware computing,hospital information systems (HISs), public displays, ubiquitouscomputing.
UBIQUITOUS or pervasive computing refers to the designof environments saturated with computation and wirelesscommunication naturally integrated to human activity . It isthis natural integration with the environment that is emphasizedin a term often used to refer to ubiquitous computing environ-ments, the disappearing computer.
Hospitals are complex information-rich environments thatinclude a significant technical and computational infrastructure,the need for coordination and collaboration among specialistswith different areas of expertise, an intense information ex-change, and the mobility of hospital staff, patients, documentsand equipment. This makes them ideal application environ-ments for pervasive or ubiquitous computing technology.
Manuscript received November 30, 2003; revised March 3, 2004. This workwas supported in part by CONACYT and UCMexus under Grants U-40799 andCN-02-60.
J. Favela and A. Preciado are with the Centro de Investigacin Cientfica y deEducacin Superior (CICESE), Ensenada 22860, Mexico (e-mail: email@example.com; firstname.lastname@example.org).
M. Rodrguez is with the Centro de Investigacin Cientfica y de EducacinSuperior (CICESE) Research Center, Ensenada 22860, Mexico, and also withthe Universidad Autnoma de Baja California (UABC), Mexicali 21280,Mexico (e-mail: email@example.com).
V. M. Gonzlez is with the University of California, Irvine, CA 92717 USA(e-mail: firstname.lastname@example.org).
Digital Object Identifier 10.1109/TITB.2004.834391
The effectiveness of information artifacts within hospitals ishighly dependent on their location, but also on being able to pro-vide adequate information to the reader. It has been pointed outthat due to their different professional background, hospitalspersonnel is likely to experience communication problems or todefine and agree on what is the most useful way to representinformation contained in artifacts . For instance, physiciansand nurses can pull different bits of data from a patients recordin order to do their work (e.g., a diagnostic versus the admin-istration of medications). Therefore, in order to be effective, aninformation artifact has to be elastic enough to provide with dif-ferent levels of representations for each possible reader, in sucha way that it results meaningful for all of them.
One of the most active areas of research in ubiquitous com-puting is the use of large displays, either as boards hung onwalls, or interactive tables used for information access, man-agement, and sharing. It has been argued that the full potentialof these devices will only emerge if they are embedded in ubiq-uitous computing environments which could sustain realistic in-teractive use . Their use in support of small groups, to pro-vide awareness of users activities and promote collaboration,has also been addressed .
In this work, we explore the use of interactive public displaysin hospitals. We were inspired by ethnographic work that iden-tified the crucial role of whiteboards hung on walls for coordi-nation , , . Public displays by themselves would offeronly limited services, thus, we explore their integration into acontext-aware mobile computing infrastructure  developedspecifically for hospitals.
In Section II, we describe a context-aware hospital informa-tion system (CHIS). Section III presents a case study aimed atunderstanding current uses of public boards. In Section IV, wediscuss the findings from the study, including a scenario en-visioned to design interactive applications for public displays.Section V describes how the displays are integrated in themobile hospital information system (HIS). Finally, Section VIpresents the results of the systems evaluation conducted withhospital personnel and Section VII our conclusions.
II. MOBILE ACCESS TO ELECTRONIC PATIENT RECORDS
Electronic patient records (EPR) integrate patient and clini-cian data so that authorized medical personnel can retrieve nec-essary information to provide adequate treatment. A recent trendin the development of these systems is the support for mobilecomputing devices that can be used by the hospital staff to ac-cess clinical records anywhere and anytime.
One such system is the CHIS , which provides access to in-formation based on the users context, and allows users to send
1089-7771/04$20.00 2004 IEEE
280 IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 8, NO. 3, SEPTEMBER 2004
Fig. 1. Context-aware client indicating the presence and location of users as(a) a list or (b) a map.
messages that will be delivered based on the recipients locationand role, or the status of artifacts such as results of analysis.The design of CHIS was based on a set of findings gatheredduring a workplace study conducted in a public hospital. Fromthis study, we identified that the medical staff need to locate rel-evant documents, such as patients records, laboratory results,and forms to be filled, locate patients and colleagues, and locateand track the availability of devices, such as medical equipmentand beds, which are moved within the hospital. Through inter-views and questionnaires, hospital staff emphasized the impor-tance of locating peers and resources, that communication ex-changes depend on context, and their belief that handheld com-puters are appropriate mechanisms to access medical data. InCHIS, a handheld computer allows medical staff to locate pa-tients, other staff members, and resources. It integrates a loca-tion estimation mechanism through which the system decideswhat information to deliver to the user. For instance, access topatients records is most relevant when near the patients bed.
A. Architecture of the CHIS SystemThe architecture of the context-aware system described above
includes the following components.Context-Aware Client: Physicians and nurses carry a hand-
held computer that estimates their location, provides themwith information relevant to their location, and allows themto fill requests and communicate with other members of thestaff. The interface of the context-aware client is based on theinstant messaging (IM) paradigm, through which users arenotified of the availability of other users and their location. Thisinformation is displayed in the form of a list (as in traditionalIM system) [Fig. 1(a)] or in a map of the area surrounding theuser [Fig. 1(b)]. The context-aware client provides access tothe HIS. In addition, the client includes a location-estimationagent that estimates the users position.
HIS: This system manages and stores the patients clinicalrecords and other data relevant to the hospital, such as whichpatients are in which beds.
HIS Agent: This agent acts as proxy of the HIS, in pro-viding access to information contained in it, and monitoring itschanges. Rules are used to indicate what type of informationshould be delivered to a user given his current location androle. For instance, when the HIS agent becomes aware that thephysician is near a patient and that this patients lab results are
available, it notifies this to the doctors client. This agent runsas a daemon on a computing device with connectivity to anagent directory and the IM server.
Location-Estimation Agent: This has a reasoning com-ponent which wraps a back propagation neural network,previously trained to map the signal strength obtained fromeach access point in the vicinity, to the users location.
Agent Broker: This handles communication between agents,which represent users, services, and devices. Information iscommunicated through extended markup language (XML)messages. To implement this service, we used the Jabberopen-source IM server (www.jabber.org) and extended itsextensible messaging and presence protocol (XMPP). Thisserver also stores the state of people and agents and notifiestheir changes to other agents subscribed to them.
Context-Aware Agent: All messages that depend on contex-tual variables for their delivery are transferred to this agent,which monitors the environment to determine whether condi-tions are met for the delivery of a contextual message.
The agents in CHIS were developed using SALSA, a classframework for implementing autonomous agents that act on be-half of users, represent services, or wrap a systems function-ality . SALSA agents communicate through an XML pro-tocol that extends XMPP.
The system was evaluated by 28 hospital staff members topredict user acceptance and obtain additional insights to en-hance the system. We used the technology acceptance model(TAM)  to estimate the medical staffs perception with regardto utility, ease of use, and intention of use. The results indicatethat 91% of the participants would use the system, 84% believedthat the use of the system would enhance their job performance,and 78% perceived the system to be easy to use. In addition,the subjects suggested new scenarios of use and proposed ex-tensions to the systems functionality. In particular, physicianswere interested in being able to visualize the results of clinicalanalysis as they become available, including X-ray images.
A discussion among the subjects that participated in the eval-uation made it clear that visualizing medical images through ahandheld computer had significant limitations. This suggestedto us the use of Tablet computers or public displays. We de-cided to further explore the second option, considering that hos-pitals already have numerous whiteboards used for coordinationand communication. We undertook a second case study, this onedirected specifically at understanding the current uses of thistype of artifacts, including whiteboards, corkboards, and X-rayviewers.
III. CASE STUDY
We conducted a workplace study to help us envision innova-tive uses of interactive public displays in a hospital. We useda combination of qualitative methodologies to understand howboards are currently used to support daily work. In particular, weused participant observation  and informal and semistruc-tured interviews with key informants . Our earlier study con-ducted in the same hospital gave us an understanding of workingpractices and procedures that facilitated the identification offindings to inform our design.
FAVELA et al.: INTEGRATING CONTEXT-AWARE PUBLIC DISPLAYS INTO A MOBILE HIS 281
A. The SettingThe study was conducted at the IMSS General Hospital in
Ensenada, Mexico. This is a public health institution providingmedical services for a potential population of 175 000 inhabi-tants. Around 800 persons work in this location. Every year theyhave an average of 302 000 medical consults, 92 000 urgent ser-vices, 3000 births, and 600 000 lab studies. With more than 130beds, the hospital covers the demands of around 82% of the pop-ulation in the area.
B. Workplace StudyFor a period of two months, we studied the area of Trauma-
tology. The study was conducted in two phases. During the firstfour weeks, interviews and observations were performed to ob-tain a general understanding of the work processes followed bynurses, physicians, lab staff, and social workers. Once an initialunderstanding of the overall process was obtained, four addi-tional weeks were spent focusing on the support provided bywhiteboards and X-ray viewers to hospital staff.
Nineteen (19) informants were interviewed, including four(4) physicians, four (4) resident physicians, three (3) nurses,three (3) assistants, four (4) directors of services, and one (1)chemist working in the laboratory. Given the nature of theirwork, we had to restrict the time of the interviews to one (1)hour with each informant, and in some cases, we had to conductseveral short interviews with the same person.
We were allowed to talk with any member of the staff whowas available to talk with us and move around the hospital byourselves or following a physician or a nurse. In this way, weconducted many informal interviews that let us to become fa-miliar with the setting, people, and their practices.
In order to gain a detailed understanding of the ways in whichindividuals use display artifacts to publish and consume infor-mation, we shadowed two resident physicians, each for two fulldays of work. Our technique is similar to that employed in otherstudies (e.g., ): We followed our informant and made notesof her activities, the artifacts that she used, and the people withwhom she interacted. Additionally, we monitored the use of oneof the whiteboards used to indicate the location of patients andthe physicians assigned to them. For a period of one week, inseven observation sessions of 2 h each, one of the researchersmonitored the use of the whiteboard. If during the period of ob-servation the artifact was used, we made notes about it includingthe kind of activity supported and the roles of participants. Eachday we recorded all the observed changes made in the artifact.The observation sessions were distributed across the day to un-derstand how people working at different hours used the arti-fact. Through this, we were able to get a better sense of whoupdated the information and how often, and who used this infor-mation and for what purposes. We were also able to determinethe coding schemes used to publish the information.
The last two weeks of the study were intertwined with thedesign phase. While design scenarios were being developed, is-sues were raised by the designers, which often led to one of usgoing back the next day to the hospital to confirm our assump-tions or gather a missing piece of information. The scenarios,thus, combine our vision of the support that can be provided by
interactive public displays with actual, and in some cases veryconcrete, issues faced every day by hospital workers.
IV. FINDINGS FROM THE STUDYThe findings from the study were used to inspire the creation
of design scenarios that depict innovative uses of interactivepublic displays in a hospital. Scenarios like this one were usedto shape the characteristics of the applications designed to runon the public display presented in Section V.
A. Scenario of Use for Interactive Public DisplaysUbiquitous computing environments offer complex technical
solutions which include technologies that are not commonlyused by the people in our study, thus, we decided to use sce-narios as a way to feedback our understandings of their prac-tices and also our vision of how their work could be augmentedwith context-aware tools. One such scenario is presented next.
While Dr. Garcia is checking the patient in bed 234, his per-sonal digital assistant (PDA) alerts him that a new message hasarrived. His handheld displays a hospital floor map informinghim that the X-ray results of patient in bed 225 are available.Before Dr. Garcia visits this patient, he approaches the nearestpublic display that detects the physicians presence and provideshim with a personalized view of the HIS. In particular, it shows apersonalized floor map highlighting recent additions to clinicalrecords of patients he is in charge of, messages addressed to him,and the services most relevant to his current work activities. Dr.Garcia selects the message on bed 225, which opens windowsdisplaying the patients medical record and the X-ray image re-cently taken. Aware of the context of the situation (patientsoriginal diagnosis, the fact that X-rays where just taken from thepatients hand, etc.), the system automatically opens a windowwith the hospitals medical guide that relates to the patientscurrent diagnosis, and an additional one with previous similarcases to support the physicians analysis. While Dr. Garcia isanalyzing the X-ray image, he notices on the map that a residentphysician is nearby and calls him up to show him this interestingclinical case. The resident physician notices that this is indeeda special case not considered by the medical guide and decidesto make a short note on his handheld computer by linking boththe X-ray image and the medical guide. He can use these linkslater on to study the case in more detail or discuss it with othercolleagues from any computer within the hospital.
Scenarios such as this one were generated to bridge the gapbetween current medical practice and an idealized pervasivemedical environment. They address current sources of mis-shapes or look to simplify complex tasks through the use ofubiquitous computing technology.
B. Desirable Features of Hospital Public DisplaysFrom scenarios such as the one presented above, we identified
the following aspects to be addressed by context-aware interac-tive public displays.
1) Users Location and Authentication: Hospitals are char-acterized by the mobility of the professionals that work there,that of the artifacts they use, such as clinical records or med-ical equipment, and even the patients, who are moved from one
282 IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 8, NO. 3, SEPTEMBER 2004
hospital area to another as required. When a doctor examines apatient, he needs to move to obtain the patients clinical recordsand other documents. Hospital workers also need to move tolocate information displayed in whiteboards. For instance, theschedule of patients operations for the current day is displayedon the whiteboard in the office of the chief surgeon, and theactivities and working area assigned to nurses are advertised indifferent boards throughout the hospital. Boards help to commu-nicate information regarding patients conditions and locations,and hospital staff often visits the boards to find this informa-tion. Thus, hospital workers require access to information fromanywhere within the hospital. In addition, as exemplified by thescenario, the public display should be able to recognize the useras he approaches it and give him access to relevant clinical datawithout cumbersome login procedures. Secure approaches to lo-cation-based authentication have been reported in the literature.
2) Content Adaptation and Personalization Based on Con-textual Information: Contextual information such as location,role, and identity should be taken into account to adapt andpersonalize the presentation of information to the user. For in-stance, in the above scenario, when Dr. Garcia is in front of thedisplay, the pervasive environment displays the messages ad-dressed to him and highlights the beds of the patients he is at-tending. A traumatologist attends mostly a small and well-de-fined number of patients assigned to him. He might be requiredto occasionally access the medical record of another patient, butthis is seldom the case. Thus, we prevent information overloadby personalizing his display to provide immediate access to theclinical records of those patients assigned to him. The personal-ization of information shown in the public display also helps todeal with limited screen real state. We observed, for instance,that a resident physician would attach sheets of paper to theside of the whiteboard to inform others of patients in beds bor-rowed from other areas, since there was no additional space onthe whiteboard.
3) Information Transfer Between Heterogeneous De-vices: In the hospital, users frequently transfer informationfrom public spaces to personal spaces. For instance, the chiefnurse might leave a note on a public board in order to advertisethe date of the next meeting; then, another nurse would writethis information into her personal agenda. We also observed aphysician going to a whiteboard at the beginning of his shift tofind out about new patients assigned to him and then writingthis information in the back of a folder he carried with him. Afew physicians actually carry PDAs and report using them torecord information displayed on whiteboards or corkboards.A pervasive environment furnished with devices of all scales,should support the simple and safe transfer of informationbetween devices. Some of these devices are public, such aslarge displays, while others are personal, such as PDAs. Inthe scenario, the resident physician decides to keep a personalrecord of the clinical case presented on the display. Thus, thephysician transfers the relevant information from the largedisplay (public space) to his PDA (personal space). A doctormay also want to transfer information from his PDA to a publicdisplay in order to present it and discuss it with a colleague.Information is also transferred from both devices when the
physician approaches the board and is authenticated; the displaywill use information stored in the users PDA to personalizethe applications running in the public display. For instance, inthe scenario, the message received by the physician indicatingthat the X-rays are available and the dates stored in the PDAof the physicians calendar are transferred to the display topersonalize the map and calendar applications.
4) Opportunistic Access to Relevant Informa-tion: Physicians often use information from previouscases in their decision-making. For this, they might rely onexperience or consult medical journals, but they seldom consultthe records of previous patients, to a large extent becausethey are difficult to locate. The EPR offers the opportunity toretrieve relevant medical cases with little effort from the user.To retrieve this information, the pervasive environment hasto take into account contextual information, such as, type ofclinical problem, and if the device where it will be displayedis a public display, that enables us to clearly appreciate andanalyze documents and images. The environment may alsoopportunistically display medical guides relevant to thepatients diagnosis as supporting information for doctors oreven locate and establish contact with a specialist who mightbe available. In the scenario, the display presents the hospitalsmedical guide relevant to the case they are discussing and linksto previous cases that were estimated to be relevant.
The issues just discussed were addressed by extending theCHIS system as described next.
V. INTEGRATING PUBLIC DISPLAYS IN A HOSPITAL
A. Extending CHISs ArchitectureThe architecture of the CHIS discussed in Section II was ex-
tended to integrate public displays that are context-aware, per-sonalize information to the user, allow the transfer of informa-tion to and from PDAs, and offer opportunistic access to clinicalinformation. The deployment diagram of the architecture is il-lustrated in Fig. 2. The additions to the architecture are includedin the display server node. They include three new agent compo-nents that communicate among themselves and with other com-ponents through the agent broker. This has allowed for a seam-less integration of the new components, since only minor mod-ifications were required to other components. We next describethe new components.
Public Display Agent (PD-a): This component acts asa proxy for the public display. It incorporates an IM clientthrough which it becomes aware of the status and location ofusers and other devices, and communicates with them via theagent broker.
Knowledge Management Agent (KM-a): This is responsiblefor displaying the hospital medical guide and previous casesrelevant to the case being consulted. The agent indexes the EPRsand the medical guides, it retrieves information from the recordof the patient being consulted by the user of the public display,such as the current diagnosis and type of analysis performedand displays a medical guide, if one is found to be relevant, anda list of previous clinical cases considered to be relevant. Theselinks can be used by the physician using the display to accesspreviously documented cases. At this point, we are using simple
FAVELA et al.: INTEGRATING CONTEXT-AWARE PUBLIC DISPLAYS INTO A MOBILE HIS 283
Fig. 2. Extended architecture of CHIS, with components used to incorporate public displays.
keyword matching to retrieve information. The KM-a can alsobe extended to offer preliminary computer-assisted diagnosis.
Users Proxy Agent (UP-a): This agent acts as proxy of theuser. It visually represents the user by presenting his photographin the display and it references information stored in the usersPDA. By selecting the icon with his own photo, a user opensa folder with references to documents stored on his PDA orWebDAV folder. Finally, if a user drags and drops a windowonto his photo icon it will add a link to the document containedin the window (EPR, X-ray image, medical guide, etc.) into thefolder in his PDA.
Through a sample application, we next describe how thesecomponents, and those of the CHIS architecture described ear-lier, interact to support the work of healthcare professionals.
B. Sample Application
We revisit the scenario presented in Section IV to illustratethe functionality of the extended architecture. Fig. 3 illustrateshow the systems components interact to support this scenario.As Dr. Garcia examines the patient in bed 234, the X-ray resultshe requested are included in the electronic record of his patientin bed 225. The HIS agent notifies the doctor by sending hima message through the agent broker. Dr. Garcia approaches thenearest public display when finished with his current patient,and before visiting patient 225. The doctors location, whichis constantly being tracked by the location-estimation agent onhis PDA, is notified to all users and agents in the environment,such as the context-aware agent and the PD-a. The PD-a ac-knowledges the users presence by displaying his photograph(see Fig. 4), indicating with this that the user has been loggedinto the system, and the applications on the display are person-alized for him.
In addition to his photograph, Dr. Garcia will be shown twopersonalized applications, the PublicCalendar and the Pub-licMap. The PublicCalendar displays the physicians agenda
and other public events published by the hospital or other staffmembers. The personalized entries are obtained from the usersPDA through the UP-a. The physician can drag public eventsof interest to him to his photo, thus transferring them to theagenda in his PDA.
The PublicMap application indicates the location of hospitalstaff and services available (printer, public display, etc) andhighlights the beds of patients assigned to the current user ofthe display. The PublicMap also shows messages addressed tothe user that depend on his location. This includes messagesrelated to his patients, such as additions to their electronicrecords (e.g., analysis results or a medical note).
In our scenario, Dr. Garcia wants to examine the recentlytaken X-rays and, thus, selects bed 225. The PD-a requests theelectronic patients records from the HIS agent and these arepresented to the user in two windows, one with the main pageof this patients medical records, and a second one with its latestaddition, in this case, the X-ray image of interest to the doctor.
The KM-a will also be notified that the user is consulting therecords of a particular patient. The KM-a will open the relevantmedical guide and manifest himself in the public display withan icon next to the users photo. While Dr. Garcia analyzes theX-ray image, he selects the KM-a agents icon for additionalrecommendations. In this case, links to previous relevant casesare presented.
While Dr. Garcia is analyzing the patients medical condi-tion, he notices on the map that a resident physician is nearbyand calls her up to discuss with her this clinical case. The resi-dent physician considers this case to be of particular interest anddecides to store a link to this information in her personal space.She does so by dragging the window with the X-ray image to heruser proxy agent represented by her photograph. She can add ashort note directly on her PDA. The resident physician can ac-cess this information later on to study the case in more detailor discuss it with other colleagues from any computer or publicdisplay within the hospital.
284 IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 8, NO. 3, SEPTEMBER 2004
Fig. 3. Sequence of interactions generated when the user approaches the public display and interacts with it.
Fig. 4. Public display personalized for a user. The PublicMap application offers location-dependent access to hospital information and services.
We presented the scenarios used to design our pervasivehealthcare environment to members of the medical staff of thesame hospital where we conducted the case study. We createda 3-min animation presenting the scenario described above and
showed it to 11 physicians and 2 nurses. Afterwards, we askedthe subjects to fill out a questionnaire and conducted briefinterviews with them looking to validate our design. The ques-tionnaire included two sections; one focused on specific aspectsof the support provided by the technology and the second partwas aimed at predicting user acceptance using TAM . In the
FAVELA et al.: INTEGRATING CONTEXT-AWARE PUBLIC DISPLAYS INTO A MOBILE HIS 285
interviews, we asked them whether the scenarios were realisticand if the suggested use of the technology would enhance theirprofessional practice.
The designs were motivated by field observations indicatingthat hospital staff often consults clinical records and other docu-ments with colleagues. Indeed, the results of the questionnairesindicate that 70% of the subjects consult clinical records or labresults with other doctors and nurses at least once a day, and allof them share this information at least once a week.
When asked what media they considered to be the most ap-propriate to consult the patients medical record, eight of them(61%) favored PDAs and seven (53%) public displays. Only tworespondents marked PCs as their preferred option and none ofthem selected the current paper versions. Answers do not sum100% since subjects were allowed to mark more than one op-tion. Responses changed slightly when asked about laboratoryresults, including X-rays. In this case, public displays were thepreferred media of nine people (69%) and seven of them men-tioned PDAs (53%). Two people selected PCs and two morepaper versions. Clearly, physicians are attracted to the idea ofhaving computerized access to patient records and favor the mo-bility offered by PDAs for clinical records and large displays forlab results because of the limited screen size of handhelds. Yet,it seems that the need to share information with other peoplefavored large displays over traditional PCs. One physician ex-pressed concern for having only a few public displays in the hos-pital; he believed that they would be insufficient to accommo-date all potential users, since physicians often start their roundsand consult clinical records at around the same time. While wehave emphasized the use of handhelds and large displays in ourdesign, we expect PCs to be part of the computing infrastruc-ture. As made clear by this comment, these devices should beadequately distributed throughout the hospital to facilitate ac-cess.
We investigated the use of medical guides by asking the sub-jects if they find them to be helpful (38% agreed) and if the tech-nology described in the scenario would make these guides moreuseful and increase their use (92% agreed). The 54% increase isan indication of a potential benefit of our design.
Finally, when asked to compare the use of public displays, asillustrated in the scenario, with that of the whiteboards they cur-rently use, 10 of the 13 respondents agreed that the functionalitydisplayed in the scenario covered all the functions of the white-boards, and the same number of people thought that there wouldbe no need for whiteboards at the hospital if the public displayswere introduced there. During the interview, one person men-tioned as a limitation of the display the fact that the map showsa single floor at a time (see Fig. 4), while on the board, theyinclude beds from different floors that are assigned to the area.This has made us consider changes in the interface of the Pub-licDisplay application to account for this.
The second part of the questionnaire aimed at predicting useracceptance of the technology. We asked 14 questions using aseven-point Likert scale with anchors ranging from stronglydisagree to strongly agree, to measure the subjects perceptionof ease of use (seven questions), perceived usefulness (fivequestions), and intention of use (two questions). The physiciansand nurses found the system to be easy to use (6.4), useful to
support clinical work (6.3), and expressed interest in using thesystem (6.7). None of them had used a large display before andtwo of them mentioned during the interviews that scenariosmade it look as if the applications were rather easy to use butthey would need to use the applications before making strongerclaims about this aspect. Nevertheless, it has been arguedthat for professionals, and physicians in particular, perceivedease of use might not be a good predictor for intention of usesince professionals should be able to quickly assimilate newtechnology that is perceived to be useful .
In the interviews, all participants found the scenario to be re-alistic and we got a clear sense that immediate and easy access tothe medical records was the most attractive feature of the tech-nology, and that they saw the combination of handhelds andpublic displays as providing adequate support to achieve this.Furthermore, our interviewees suggested extending the systemto support distributed collaboration while working on the publicdisplay. In particular, one physician suggested to be able to lo-cate colleagues in other hospitals with whom they have workedbefore and at times call to consult a clinical case. We perceivethat the support could include providing awareness of the pres-ence of remote contacts, application sharing, and audioconfer-encing.
Privacy is an important issue in the use of large public dis-plays . Given the uses we propose for them and their lo-cation within a hospital, our displays could be considered to besemi-public, which minimizes privacy concerns . These de-vices are to be located in areas restricted to hospital staff wherethe medical records are currently stored, thus, access will besimilar to that of paper-based clinical records. Other informa-tion such as the schedule of hospital staff is also semipublic andcurrently displayed on corkboards.
Hospitals are complex work environments where people andinformation are distributed, thus demanding considerable coor-dination and communication among the professionals that workin such settings. Electronic medical records are an importantstep toward providing adequate access to clinical information.However, the most precious resource is the attention of the med-ical personnel. With adequate support to estimate the context ofwork, context-aware systems can deliver information that is rel-evant to the users location, identity, and/or role.
We have presented our efforts to extend a mobile HIS with in-teractive public displays as a direction toward the developmentof a pervasive hospital environment. Our efforts included con-ducting a workplace study in a hospital to understand currentuses and limitations of public artifacts such as whiteboards andcorkboards from which design scenarios and insights were de-rived. These scenarios formed the basis for our design.
An evaluation with hospital staff provided preliminary evi-dence that the functionality offered by the system would helpaddress issues faced by healthcare service providers. We plan toconduct a pilot within the hospital by bringing a large display tothe premises and have hospital staff work with the applicationsdeveloped with it.
286 IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE, VOL. 8, NO. 3, SEPTEMBER 2004
The authors would like to thank the medical staff at IMSSGeneral Hospital in Ensenada, Mexico, for their help in evalu-ating the prototype described in the paper.
REFERENCES J. Bardram, R. E. Kjr, and M. Pedersen, Context-aware user authen-
tication Supporting proximity-based login in pervasive computing, inProc. UBICOMP, 2003, pp. 107123.
 J. Bardram and C. Bossen, Moving to get ahead. Local mobility andcollaborative work, in Proc. ECSCW, 2003, pp. 355374.
 C. Bossen, The parameters of common information spaces: The hetero-geneity of cooperative work at a hospital ward, in Proc. CSCW, 2002,pp. 176185.
 F. D. Davis, Perceived usefulness, perceived ease of use, and user ac-ceptance of information technology, MIS Quart., vol. 13, no. 3, pp.319340, 1989.
 P. J. Hu, P. Chau, O. Liu, and K. Y. , TAM, examining the technology ac-ceptance model using physician acceptance of telemedicine technology,J. Manage. Inf. Syst., vol. 16, no. 2, pp. 91112, 1999.
 E. M. Huang and E. D. Mynatt, Semi-public displays for small, co-lo-cated groups, in Proc. CHI, 2003, pp. 4956.
 B. Johanson, A. Fox, and T. Winograd, The interactive workspacesproject: Experiences with ubiquitous computing rooms, IEEE Perva-sive Comput., vol. 1, pp. 7178, Apr./June 2002.
 G. McCracken, The long interview, in Qualitative Research MethodsSeries, M. L. Miller, Ed. Newbury Park, CA: Sage, 1988, vol. 13.
 M. Muoz, M. Rodriguez, J. Favela, V. M. Gonzalez, and A. I. Mar-tinez-Garcia, Context-aware mobile communication in hospitals,IEEE Computer, vol. 36, pp. 6067, Sept. 2003.
 L. A. Perlow, The time famine: Toward a sociology of work time, Ad-minist. Sci. Quart., vol. 44, pp. 5781, 1999.
 M. Reddy and P. Dourish, Coordination in heterogeneous work: Infor-mation and representation in medical care, in Proc. ECSCW, 2001, pp.239258.
 M. Rodrguez and J. Favela, Autonomous agents to support interop-erability and physical integration in pervasive environments, in Proc.AWIC, 2003, pp. 278287.
 J. Spradley, Participant Observation. New York: Thomson, 1997. D. S. Tan and M. Czerwinski, Information voyeurism: Social impact
of physically large displays on information privacy, in Proc. CHI 2003,2003, pp. 748749.
 M. Weiser, The computer for the twenty-first century, Sci. Amer., pp.94104, Sept. 1991.
Jesus Favela received the B.S. degree in civil engi-neering from the National Autonomous University ofMexico (UNAM) and the M.Sc. and Ph.D. degrees incomputer aided engineering from the MassachusettsInstitute of Technology, Cambridge.
He is a Professor of Computer Science at theCenter of Scientific Research and Higher Educationof Ensenada (CICESE), where he leads the Collabo-rative Systems Laboratory and heads the Departmentof Computer Science. His research interests includecomputer-supported cooperative work, ubiquitous
computing, and information retrieval.Dr. Favela is a Member of the ACM, IEEE-Computer Society, and current
President of the Sociedad Mexicana de Ciencia de la Computacion (SMCC).
Marcela Rodrguez received the M.Sc. degree incomputer science from the Center of Scientific Re-search and Higher Education of Ensenada (CICESE)and the B.Sc. in computer engineering from theAutonomous University of Baja California (UABC).She is working toward the Ph.D. in computer scienceat CICESE.
She is a Lecturer in Computer Engineering atUABC. Her research interests include ubiquitouscomputing, autonomous agents, and computer-sup-ported cooperative work. She is a Student Member
of the ACM.
Alfredo Preciado received the B.Sc. degree in com-puter engineering from the Universidad de Guadala-jara. He is a graduate student in computer science atthe Center of Scientific Research and Higher Educa-tion of Ensenada (CICESE).
His research focus is on medical informatics, ubiq-uitous computing, and computer-supported coopera-tive work.
Vctor M. Gonzlez received the B.S. degree inelectronic communication from the Institute ofTechnology of Monterrey (ITESM), the M.S. degreein information and computer science from theUniversity of California, Irvine, and the M.S. degreein telecommunication and information systems fromthe University of Essex, U.K. He is working towardthe Ph.D. degree in information and computerscience at the University of California, Irvine.
He conducts empirical research on informationtechnology usage to support activity management.
His studies have included both office and home settings. Currently he isstudying the strategies that information workers use to manage multipleprojects under fast-paced conditions.
tocIntegrating Context-Aware Public Displays Into a Mobile HospitalJesus Favela, Marcela Rodrguez, Alfredo Preciado, and Victor M.I. I NTRODUCTIONII. M OBILE A CCESS TO E LECTRONIC P ATIENT R ECORDS
Fig.1. Context-aware client indicating the presence and locatioA. Architecture of the CHIS SystemContext-Aware Client: Physicians and nurses carry a handheld comHIS: This system manages and stores the patient's clinical recorHIS Agent: This agent acts as proxy of the HIS, in providing accLocation-Estimation Agent: This has a reasoning component which Agent Broker: This handles communication between agents, which rContext-Aware Agent: All messages that depend on contextual vari
III. C ASE S TUDYA. The SettingB. Workplace Study
IV. F INDINGS F ROM THE S TUDYA. Scenario of Use for Interactive Public DisplaysB. Desirable Features of Hospital Public Displays1) User's Location and Authentication: Hospitals are characteriz2) Content Adaptation and Personalization Based on Contextual In3) Information Transfer Between Heterogeneous Devices: In the ho4) Opportunistic Access to Relevant Information: Physicians ofte
V. I NTEGRATING P UBLIC D ISPLAYS IN A H OSPITALA. Extending CHISs ArchitecturePublic Display Agent (PD-a): This component acts as a proxy for Knowledge Management Agent (KM-a): This is responsible for displ
Fig.2. Extended architecture of CHIS, with components used to iUser's Proxy Agent (UP-a): This agent acts as proxy of the user.B. Sample Application
Fig.3. Sequence of interactions generated when the user approacFig.4. Public display personalized for a user. The PublicMap apVI. R ESULTSVII. C ONCLUSIONJ. Bardram, R. E. Kjr, and M. Pedersen, Context-aware user auJ. Bardram and C. Bossen, Moving to get ahead. Local mobility anC. Bossen, The parameters of common information spaces: The heteF. D. Davis, Perceived usefulness, perceived ease of use, and usP. J. Hu, P. Chau, O. Liu, and K. Y. , TAM, examining the technoE. M. Huang and E. D. Mynatt, Semi-public displays for small, coB. Johanson, A. Fox, and T. Winograd, The interactive workspacesG. McCracken, The long interview, in Qualitative Research MethodM. Muoz, M. Rodriguez, J. Favela, V. M. Gonzalez, and A. I. MarL. A. Perlow, The time famine: Toward a sociology of work time, M. Reddy and P. Dourish, Coordination in heterogeneous work: InfM. Rodrguez and J. Favela, Autonomous agents to support interopJ. Spradley, Participant Observation . New York: Thomson, 1997.D. S. Tan and M. Czerwinski, Information voyeurism: Social impacM. Weiser, The computer for the twenty-first century, Sci. Amer.