SECTION ON COMPUTERS AND OTHER TECHNOLOGIES

(1)

SECTION

ON

COMPUTERS AND OTHER TECHNOLOGIES

SUNDAY,OCTOBER 10 2:00 pm-5:00 pm

Franklin Square, TheGrand Hyatt Washington, DC

1:00pm Welcome

S. Andrew Spooner, MD, FAAP Chairman, ScientificCommittee

1) 2:00 pm Clictate: A Structured Reporting Tool for Guideline-based Care

Kevin B.Johnson, MD, MS, FAAP, Aletha Ak-ers,Janet R. Serwint MD. Division of Pediat-rics, The Johns Hopkins University School of Medicine, Baltimore,MD.

2) 2:20pm Computerized Voice Recognition as a Re-placement for Medical Transcription StephenM.Borowitz, MD.Department of Pe-diatrics, University of Virginia, Charlottes-ville, VA.

3) 2:40 pm Instant Pocket Wireless Telemedicine Consultations

Loren G. Yamamoto, MD, MPH, MBA, FAAP. Department of Pediatrics, University of Hawaii John A. Bums School ofMedicine, Honolulu,HI.

4) 3:00pm WWW-based Applications for the Neona-tal Intensive Care Unit

Charles Hu and Ray Duncan, MD, FAAP. Medical College ofWisconsin, Madison, WI, and Cedars-Sinai Medical Center, Los An-geles,CA.

5) 3:20 pm Introducing a Bedside Terminal System in theNICU: AssessmentofUserSatisfaction RichardJ. Powers, MD, FAAP, and Jeanette Asselin, RRT, MS. Children's Hospital Oak-land,Oakland, CA.

3:40-4:00pm Coffee Break

6) 4:00pm Information Seeking Behaviors Among PediatricResidents

Jay D. Eisenberg, MD. Pediatric Pulmonary Division, Dept.of Pediatrics, Oregon Health Sciences University, Portland, OR. (Spon-soredbyDana AVBraner,MD)

Handheld Computers in Residency Train-ing: Development of a Mobile Medical Records System

JeromeK. Wang,M.D. and Ray Duncan,MD, FAAP. Departments of Intemal Medicine, Pediatrics, and Enterprise Information Ser-vices,Cedars-Sinai MedicalCenter, Los An-geles,CA.

8) 4:40pm ANew NaturalLanguage Processing Tool for CaseSimulations

Christoph U. Lehmann, MD, CFAAP, Bach Nguyen, GeorgeR.Kim, MD, FAAP, Kevin B. Johnson, MD,MS,FAAP, and Harold P.

Leh-mann, MD, PhD,FAAP. Department of Pe-diatrics, Johns Hopkins University, Balti-more,MD.

5:00pm Adjourn

ABSTRACTS

1

CLICTATE: A STRUCTURED REPORTING TOOL FOR GUIDELINE-BASED CARE

Kevin B.Johnson, MD, MS,FAAP, Aletha Akers, Janet R. Ser-wint MD. Division of Pediatrics, The Johns Hopkins University SchoolofMedicine, Baltimore, MD.

Background: The recent emphasis on the critical need for com-puter-based patientrecordsincludes an assumption that clinicians will usethese systems to document patient care during the patient encounter. In most ambulatory and inpatient settings, this goal has not yet been realized. In conjunction with developing a com-puter-based patient record (CPR) for our academic medical center, we surveyed our residents about their preferred approach for entering data into the computer. They recognized a need for efficient, flexible documentation tools that used point-and-click dataentrywhenever possible. In response to their interest and our desire for on-line encounter summaries, we developed a struc-tured reporting environment. This reportdescribesthefeaturesof thatsystem, and an evaluation of how it has been used to date.

Methods: Over a 1 yearperiod, focus groups of residents and primarycare preceptors met to identify the goalsof the program andthecontentsof the health-maintenance templates. During that time, thefirstversion of theprogram, Clictate, was created and tested.Attheend of theyear, 3 versionsofthe programhad been created. After approvalwasobtainedfromourclinicaland med-ical recordleadership, the program was implemented in a pedi-atriccontinuityclinic. Wethen allowedthe program to be used for 1 year, after which the templates were revised, and additional enhancements weremade tothe program. Residents then were enrolledin alargestudy evaluating howClictatewasused andto what extent ithadan impact onpatient-physician interaction. As a partof this study, residentswere asked tocomplete a survey describing howthey used Clictateattheconclusion ofasubsetof theirencounters.Thesubsetincludedone visitfrom eachpatient undertheageof18months that came totheclinicfor "well-child care"duringthe studyperiod.

Results:There were 240 patient enrolled in thestudy. Residents retumed 229, or 95% of the surveys. The enrolled visits were uniformly distributed across all ages (through 15 months) at which health-maintenance visits occur. Using a 5 point Likert scale, 75% of theresidentsfound thedevelopmental milestones sec-tion tobe "a greatdealof help (5)"or "somehelp (4),"and 73%of theresidents rated the anticipatory guidancesectionsimilarly.Other sectionsthatreceived a 4 or 5were thephysicalexam(51%) and overall assessment(42.8%)sections. Of those4sections,the antici-patory guidance section required typing in addition to mouse selectioninalmost allcases.Thesectionfor socialhistory,asection thatrequired almost exclusivelytypewrittendataentry,received the lowestscorefor usefulness, with46.5%of the residentsstating thatit waseitherof"nohelpatall"or"didn't usesection." When asked wheninthecourseof the day they completed eachsection of documentation,morethan43%of the residents used the com-puterduringthevisitfor theinterimand social history,and more than 54% used it for anticipatory guidance and developmental assessment. However, over 74% of the residents documented physical exam, assessmentandplans before seeingthe next pa-tient, orsometimethereafter,with mostresidents completingthis documentation "attheend of the clinic session." Residentuseof Clictateduring the clinicsession wascorrelated with the degree of help they thoughtthe templateprovided (R=0.57; P<0.01).

Conclusion: Residents find many aspectsof aguideline-based

(2)

influence or be related to the timing of their documentation. Further research is needed to improve the ease with which all documentation can be completed by the end of a patient encoun-ter.

2

COMPUTERIZED VOICERECOGNITION AS A REPLACEMENT FOR MEDICAL TRANSCRIPTION

Stephen M. Borowitz, M.D., Department of Pediatrics, University of Virginia,Charlottesville, Virginia 22908

Background: Dictation with transcription is widely used to im-prove the completeness and legibility of patient records and as-surecompliance withfederal documentation requirements. Tran-scriptioncosts are$0.17-$0.25 per line and mayamount to11%of collected fees for outpatient visits. Computerized voice recogni-tion is a meansof decreasing the cost and improving the turn-around of transcribed documents. Little research has been con-ducted examining the utilityof computerized speech recognition dictationinmedical practice.

Methods: The author kept track of thetimespentdictating and editing outpatient notes with a stopwatch and log book. Dictations were randomly assigned to human or computer transcription. Computer transcription was performed by eitherIBM ViaVoice software or DragonNaturally Speaking software on a Dell 266 mHz Pentium computer withanAWE64 sound card. Data col-lected included (1) date of visit, (2) dateof dictation, (3) time spent dictating, (4) time spent editing, (5) date the final note was com-pleted and filed, (6)length of the final note, and (7) type of visit (initial consultation or follow-up visit).

Human DragonNaturally Speaking

Numberof dictations 420 162

Initial/follow-up visits 156/264 54/106 Dictationtime(sec) 179.88±62.83 197.66±74.22 Editingtime(sec) 148.95±74.08 225.15±95.64 Totaldictating and 328.82±130.05 422.81 ±160.45

editing (sec)

Completiontime(days) 5.94±3.80 0.28±0.50 Length of finalnote 50.01 ±16.61 50.78±17.75

(65 characterlines)

IBM ViaVoice Pvalue

210

74/136 0.7254

177.64±64.81 0.0058

234.82±110.03 <0.0001

412.46±163.25 <0.0001

0.15 ±0.36 <0.0001

44.04±15.79 <0.0001

Results: Data are summarized in the table. During the study period, the author dictated 792 notes. 420 were completed bya transcriptionist, 210by ViaVoice, and162by NaturallySpeaking. The proportion of initial and follow-upvisits transcribedby the three systems was comparable. The time spent dictating with Naturally Speaking was significantly greater than with either human transcriptionor ViaVoice. Theamountoftimespent edit-ingwassubstantially greater using bothViaVoiceandNaturally Speaking than using human transcription. Notes generated by ViaVoice were shorter than notes generated with human tran-scriptionorNaturallySpeakinghowever therewas nodifference inthelength ofnotesgeneratedbyNaturally Speakingand human transcription. The time to the completion ofnotes was

signifi-cantlyshorter using both ViaVoice andNaturallySpeaking than usingahumantranscriptionist.

Conclusions: Currently available computerized dictation sys-tems are practical and fast enough to replace human transcription. The time spent dictating and editing notes with computerized voice recognition is approximately 25% greater than if the notes are transcribed by a medical transcriptionist however notes are completed substantially faster when generated by computer as compared to a humantranscriptionist. With continuing techno-logical improvements, the incentives to use computerized voice recognition systems in medical practice will far outweigh any of itsdrawbacks.

3

INSTANT POCKET WIRELESS TELEMEDICINE CONSULTATIONS

Loren G. Yamamoto, MD, MPH, MBA, FAAP Emergency Ser-vices, Kapiolani Medical Center For Women And Children De-partmentof Pediatrics, University of Hawaii JohnA. BumsSchool ofMedicineHonolulu,Hawaii

Background: Consultation with specialists such as neurosur-geons may not be immediatelyavailableinmanynon-tertiarycare settings. Stat neurosurgicalconsultation is often necessaryinthe acutely head injured child tofacilitate transfer and to prepare an operative teamatthe tertiarycenter.Mostclinicalinformationcan begiven to the neurosurgeon over thephone, but the patient's CT scan is best viewedby the neurosurgeon rather than described verbally. Teleradiology can transmit the patient's CT scan to a receiving teleradiology unit, but the neurosurgeon must travelto theteleradiology unitto viewthe image. Theavailability of wallet sized pocket computers and wireless modem card devices can provide a neurosurgeon with pocketteleradiology capability.In addition tocellular telephone networks, many new wireless dig-italnetworks are currently available toprovidewireless connec-tivityfor manydifferent purposes. The purposeof this studyis to demonstrate the feasibility of wireless telemedicine links for in-stant access toconsultants anywherethey may be.

Methods:Apocket computer(available at$699) witha 640by 240pixel screeninstalled with awireless digital modem using the CDPD (cellular digital packet data) wireless data network was used to accessCT scanimages fromawebserver ontheintemet at 19,200 baud.This computerunit measures 19.7 cmby12.7cm by3.2 cm (162 grams). Nospecial softwarewas installedonthe pocket computer. No connection to a separate phone unit was required. Five sets of testCT scan images(each setconsistedor1, 3, or 4CT cutstotalling 11 CTcuts)were uploaded into aweb serverfor download testingintothepocket computer.

Results: These five setsof CT scan images, totalling 188.7 Kbytes inJPEG (averaging 17 Kbytes per CTcut), wereaccessed wire-lessly and viewed using the pocket computer's built in web browser without problems. Nowireconnections,additional soft-ware orspecial computer commandswere requiredto viewthe images. Viewing times from turning the unit on to seeing the images took 5 minutes(the first 50 secondswererequiredtologon to the world wide web) foranaverage download speed of755 bytes per second(7,550baud), short of thetheoreticalmaximumof 19,200baud. This was done locally within thesame city, but it could justaseasily have occurred2000miles awayor more.

Conclusion: Pocket computers with wireless internetaccesshave powerful multimedia and telecommunicationcapabilities permit-ting telemedicine functions in a pocket device. These units are convenient,portable and inexpensive. Theyfacilitate the further growthanddevelopmentof telemedicine byeliminatingthe need to travel totelemedicine centers or to access a PC for connection to anetwork.

(3)

4

WWW-BASEDAPPLICATIONSFOR THE NEONATAL INTENSIVE CAREUNIT

CharlesHuand Ray Duncan, M.D., MedicalCollege of Wiscon-sin,Madison, Wisconsin,andDepartments of Pediatrics and En-terprise Information Services, Cedars-Sinai Medical Center, Los Angeles,California.

Background: Computerized decision-support applications can simplify the process of ordering drugs and solutionsinthe Neo-natal Intensive Care Unit (NICU) and reduce the incidence of calculation errors. The World Wide Web (WWW) has made it possibletorapidly build and deploy such applicationsin a

plat-form-independentmanner.

Methods: Parallel versions of several WWW-based applica-tions for the NICU were developed in both client-side JavaScript and server-side Microsoft Active Server Pages (ASP) to test performance and portability on a variety of WWW browsers. Certain of the applications were previously imple-mented in Microsoft Basic, VAX RDB, Apple HyperCard, or

Symantec ThinkC. Information is collected from the clinician viaHyperTextMarkup Language (HTML) "forms" displayed in a WWWbrowser.

Calculationsarecarriedout ontheclient(JavaScriptversions) or ontheserver(ASP versions), using information storedonthe server.The results aredelivered to theuser via the browser.

Graphsandchartsarecreateddynamicallyandembeddedin WWW pages usingServerObjects' ASPChart,anActiveX mod-ule thatcanbe calledfromanASPapplicationtogenerateJPEG images.

Results: Thefour WWW-based patient-care applications listed beloware now inroutine use in the CSMCNICU.

(1) QuickIV:Thephysiciansupplies the patient's weight and the desired amount of fluids, dextrose, sodium, potassium,

calcium,andheparinperunitweightperday.Theapplication

calculates the exact amounts of fluids, dextrose, electrolytes,

andheparintomix intheIVsolution, alongwith thedelivery rate.

(2) Quick Drip: The physician selects a constant infusion medication from a list and supplies the patient's weight, the desired volume perhour,and thedesired dose-per-weight-per-unit time. The application checks the dosage against built-in limits and calculates the amount of drug to be mixed in a standard volume.

(3)Code Sheet: The physician supplies the patient's name, weight, and birth-date. The application calculates patient-spe-cific dosages for drugs used during a neonatal resuscitation, and generatesa"Code Sheet" thatcanbeprintedandpostedat the bedside forreadyreference duringanemergency.

(4) Growth Chart: Theapplication generates three different charts ofweightandhead circumferencefor theperiodfrom24 weeks gestation to term-plus-three-months, based on curves published in the pediatric literature, and plots the patient's

weightsonthecurves.

Conclusion: Theexplosionof the Internet and theWWW,and theproliferation of low-cost, powerful, easy-to-use develop-ment tools forWWW servers and applications, have made it easytobuilduser-friendly, graphically-enabled, platform-inde-pendent decision support applications for physicians and nurses working in the NICU. These applications can be ac-cessed from clinical workstations

throughout

thepatientcare areas, aswell asthe desktop computers in thephysician and nurseoffices.

5

INTRODUCING A BEDSIDETERMINAL SYSTEM INTHE NICU-ASSESSMENT OF USER SATISFACTION

RichardJ. Powers, M.D., FAAP andJeanette Asselin, RRT,MS,

Children's Hospital Oakland,Oakland, CA.

Background: We recently introduced an electronic charting sys-tem intoan area of ourNICU, as phase I of a proposed unit-wide installation. To evaluate the system, and compare it with our conventional paper charting system, we administered a customer satisfaction survey to the users of the system at the end of the first phase.

Methods: The Hewlett Packard CareVue 9000 system was in-stalled inthe mostacute8bed room ofa 52bed tertiaryreferral NICU. AllNICU nurses and respiratory therapists were given4 hours ofclassroom training with the system; other users received on the job training after "go-live." After the "go-live" date, no paper records were used in this room. After a 15-month trial period, satisfactionwassurveyed among all usersof the system, using acustom survey tool.

Results: Overall response rate to the survey was 166/285 (58%). One hundred thirty-four questionnaires were included in this analysis. The breakdown of respondents was99nurses,27 physi-ciansand9"others." Overallsatisfaction with the electronic chart-ing systemcompared with the conventional paper system was 48% satisfied vs. 52%not satisfied or undecided. Six variables wereanalyzed for their correlation with overall satisfaction: use of a PC athome; greater than 10 years experience in the NICU; professional discipline, M.D. vs. RN or Respiratory Therapist; frequency of use,<lx/wk vs.2lx/wk; and whether the system wasusedfor retrievalof data or for other tasks, including docu-mentation, careplanning or combinations of these. Multivariate correlation analysis showed the onlysignificant factors influenc-ingsatisfaction werefrequency of use lx/wk(p<.01); and use of the system for retrievalof data alone as compared to documenta-tion ormultiple tasks (p<.01).

Conclusion: Overallusersatisfactionwas48% whencomparedto paperflowsheets. This reflects thedifficultyin introducing elec-troniccharting intoabusy NICU setting with high acuity. Two contributing factors were significant in the acceptance of this technology:frequency of useandcomplexityof the tasks required oftheuser.Acceptanceof anynewcharting systemcanbe diffi-cult. Better acceptanceof electronicchartinginthe NICU setting may beachievedbyintroducingit inless acuteareas, encouraging abroaderutilization, and insuringallusers areexposedto itona regular basis.

6

INFORMATION SEEKING BEHAVIORS AMONG PEDIATRIC RESIDENTS

Jay D. Eisenberg, MD.,Pediatric Pulmonary Division, Dept. of Pediatrics, Oregon Health Sciences University (OHSU),Portland, OR.(Sponsored byDana A.V.Braner,MD)

(4)

colleague.Differences and associations among training levels, sce-narios, and types of questions associated with each information resource were examined. Results: Seventeen residents generated 116 questions: 42% wereabout diagnosis, 30% about therapy, and 28% weregeneral questions. Information resources used to pursue the questions were: MEDLINE 40%, textbook 29%, reference texts 11%, WWW 10%, human 6%, and journal articles 4%. 44% of questions asked byfirst year residents (PL1) were about diagnosis compared to 28% of questions asked by second year (PL2) and 27%of questions asked by third year (PL3) residents. These dif-ferences were not significant (p>.05). Different information re-sources were usedby residents atdifferent training levels, with 44% of PL1 residents using textbooks as their most common resource,whilePL2and PL3 residents primarily used MEDLINE (53% and 40%, respectively) (p<.005). MEDLINE was the most commonresourcefor questions about therapy and general disease topics.Textbookswerethemost commonresourcefor questions about diagnosis. There wasasignificant association between type of question and informationresourceused(p<.001). Residents felt that textbooks lagged behind the primary literature but were usefulfor guidelines for diagnosis. They preferredreviewarticles to primary research articles. Some residents were not sure if MEDLINEprovided the type of resource they wanted (e.g.,NIH management guidelines). Most residents knew about WWW search engines but didnotknow advanced searching techniques. Evidence-based medicine resources were not familiar to mostof the residents. Conclusion: With more experience, residents ask more questionsabout therapy andfewer questions about diagno-sis.Information management skills should include advanced bib-liographic database andWWWsearching, and accessing and eval-uating evidence-based medicine resources. Textbooks are a common information resource and their useshould notbe de-emphasized. Furtherassessments of informationseeking behav-iors utilizingmultiple techniques inmultiple settings should be performed as medical informatics curriculaaredeveloped.

7

HANDHELD COMPUTERS INRESIDENCY TRAINING: DEVELOPMENT OFAMOBILEMEDICAL RECORDS SYSTEM

Jerome K.Wang,M.D.and Ray Duncan, M.D.,Departments of

InternalMedicine, Pediatrics, and EnterpriseInformation Services, Cedars-Sinai MedicalCenter,LosAngeles, California.

Background: Theuseof mobile computing devices among phy-sicianshasbecomeincreasingly widespread. In ourmedical cen-ter, manyphysicians-in-training rely onportable computers for organization ofdaily tasks, electronic mail, quick access tokey medicalinformation, and personalorwork-relatednotes. Awide varietyofpocket-sized computing devicesareavailable, each with distinctadvantages and limitations. We describe thedevelopment

of a prototypic mobile database that allows its users to more efficiently document, communicate, andaccessimportantpatient information. In the future, this concept may play a role in the transitionbetweenapaper-based and electronic medical recordin ourhealth system.

Methods: After assessing the currently available devices and development tools,wechosetobaseourapplicationonthe Win-dows CE operating system(MicrosoftCorp.). Thegraphical user-interface and theunderlying database tables for the mobile device wereimplemented using Visual CE (Syware, Inc.),aWindows CE

development tool that runs on the Windows 98 platform. The

completed applicationwasthen downloadedto ahandheld Win-dows CEdevice for mobileuseandwaspilotedinthe Medicine-Pediatricshouse-staff continuity clinic for electronic documenta-tionof patientencounters.The informationresidingonthe mobile computerwassynchronized ona regularbasis witha Microsoft Access 97database on a desktop PC. The printed reports were generated from the Access database, including complete

docu-mentation of the patient encounter, prescriptions, and problem lists. Furthermore, the mobile database informationwas reorga-nized into a more efficient relational database structure in Mi-crosoft Access,inanticipationof eventualuploadsinto a hospital-wide clinical data repository.

Results:Thepilotuseofamobile device databaseapplicationin thecombined Internal Medicine-Pediatricambulatory clinicinour hospital has resulted in subjective improvementsinthe quality, readability, and turn-around time for medical documentation. House-staff appreciate theabilitytohavekey patient information immediately available withoutrecourse totheprinted chart andto share thisinformation among members of the patientcare team. Objective measurements of impact on easeof access topatient information, physician communication, and patient satisfaction arestillbeing collected.

Conclusion: The use of a mobile database application in the ambulatory clinic allows the collection and retrieval of patient informationatthepoint-of-care with greaterefficiency, minimiz-ing repetitious data entry andenhancing documentation.In addi-tion,itenablessynchronizationof the patient information witha desktop- orserver-baseddatabase, and may prove tobean im-portant step in the transition toward a full-fledged computer-based medical record.

8

A NEW NATURALLANGUAGEPROCESSING TOOL FOR CASESIMULATIONS

ChristophU.Lehmann,MD, CFAAP, Bach Nguyen,GeorgeR. Kim, MD, FAAP, KevinB.Johnson, MD, FAAP, and HaroldP. Lehmann, MD, PhD, FAAP., Dept ofPeds, Johns Hopkins Uni-versity,Baltimore,MD.

Background: Computer-based case simulations of patient en-counters,usingvaryingdegreesofverisimilitude,have been

rec-ognizedasapopularandbeneficial format for medical education including CME. Case simulations enable learnerstoacquire eval-uationand management skills andbecome familiar withcurrent

developmentsinmedicineinanefficient andreproducible man-ner without having to expose real patients to the learner. To

develop good history-taking skills,thetraineemustlearntoelicit responsesfrom patientsin alogical, thorough, andsensitive fash-ion. Unlike programmer prepared history menus,Natural Lan-guageProcessing(NLP)forces thelearnertoformulate the appro-priate options on their own and provides a higher degree of realism andanimprovedlearningexperience.

Methods: The Interactive Patient II (under construction at

http://omie.medjhmi.edu/IP2)is amultimedia,Web-based, clin-ical simulationdesignedtoteach medicaltraineeshistory-taking

andphysicalexaminationskills. Wedevelopedanaturallanguage

parser called General Recognition and Analysis of Sentences and Phrases (GRASP) toallow users to inquire abouthistory-related knowledge appropriate fora physician-patientrelationship. The core process of GRASP is to return a Canonical Phrase and the relatedcasefindinginresponsetoanyuserquery.Thisis accom-plished in two steps: First, word-level synonyms are identified and Canonical Termsarereturned.Second,sorted CanonicalTerm combinationsareusedtodetermine thecorrectCanonical Phrase. By allowing phrases to be linked to multiple Canonical Terms,

GRASP allows for the recognition of ambiguous words. Using

CanonicalTermcombinations,GRASPisabletoidentifythecorrect Canonical Phrase andimplicitlytodetermine theappropriate word sensefor theambiguousphrase.

Results: To evaluate GRASP, fifty random findings from the

knowledge databasewerechosen andquestions/queries formu-lated that a user might use to access these findings through

GRASP. All queries were entered into GRASP and in 48 cases

(5)

Phrase was retumed which hadahigher priorityinthe sorting of CanonicalTermcombinations. While the resulting Canonical Phrase was not anexact match, the concept retumed wassimilar enough toresultin a meaningful answer to a user's query. Six sample questions containing two ambiguous words ("arms", "drugs") wereused.In each case, GRASP wasable to identify the appro-priateCanonical Phrase. The delay between a query and the return ofaCanonical Phrase and the related finding inthe knowledge database varied significantly. Detailed and specific queries (i.e. "Does yourchest pain radiateintoyour arm?")resultedin signif-icantlyfaster (less than 2 seconds) responses than broader and

more generalizedqueries (i.e. "Do you have pain?") (3-5seconds). Conclusion: We created a naturallanguage interface for case simulations that has the ability to handle ambiguous word sensesand to match user questions/queries to unique Canon-ical Phrases andpatient findings in our knowledge database. GRASP allows learners to interact witha case simulation at a higherlevelof verisimilitude resultinginanimprovedlearning experience.

This work was funded in part by the National Library of Medicine (Grant#F38LM00064) and by theNational Boardof

(6)

1999;104;669

Pediatrics

Andrew Spooner