CASUS An authoring and learning tool that sup¬port diagnostic reasoning

Martin Fischer, Ludwig-Maximilians-University Munich, Germany CASUS: An authoring and learning tool that support diagnostic reasoning Summary Zusammenfassung folgt folgt Introduction – Why CASUS? Over the last years a number of multimedia authoring systems evolved for off-line and, more recently, also for online presentation of interactive educational software Nevertheless, all of the available software packages require technical skills including scripting for advanced features on top of a pedagogical concept, graphical artwork and multimedia content data These requirements limit the number of content experts as potential authors not only in medicine In consequence, multimedia productions require interdisciplinary costly teams including software developers, instructional psychologists and graphical artists in addition to the content experts and the audiovisual content data On the one hand this dilemma led to a limited number of high quality multimedia titles in medicine covering all mentioned areas of expertise – “hand-crafted” pieces of artwork On the other hand several customized medical authoring environments were developed to allow a more efficient compilation of well-structured medical content This approach resembles an assembly line and restricts the freedom of the authors significantly Despite of all the efforts to make authoring as easy as possible, most systems were unable to generate a critical mass of content It was shown that medical students in their clinical education in Germany predominantly collect data and are mostly unable to apply their knowledge to a hypothetical deductive clinical reasoning process (Gräsel & Mandl 1993) We concluded that these difficulties were at least in part due to too little exposure to relevant clinical problems There was no clearly defined list of key clinical problems to be covered by each student available at the university of Munich ZSfHD 3/99 In light of the portrayed educational problems and the ambivalent technical situation we decided to create another authoring system - CASUS - that allows clinical content experts to quickly create well structured case studies without programming skills These case studies were intended to round up bedside teaching by enhancing the clinical reasoning processes for all relevant clinical symptoms and syndromes Peer reviewed case collections were envisoned for all medical students nationwide covering the whole clinical curriculum This was in 1994 This paper describes the concept of the CASUS-authoring system with a special focus on the support of the clinical reasoning process and evaluation data on authoring and learning with CASUS The concept of CASUS Constructivist approaches to learning and teaching emphasize two characteristics of learning environments which should impart useable knowledge (e.g Cognition and Technology Group at Vanderbilt 1992; Collins et al., 1989): (1) Authenticity: The learning problems used should be similar to real problems in their central characteristics (2) Complexity: The complexity of the problems should not be reduced too much, and above all should be presented embedded in an authentic context In order to achieve an appropriate level of authenticity, it is requested of the CASUS authors that they include diverse multimedial elements (e.g audio, pictures, and video) while designing their cases Also, the authors are encouraged to use as many interactive elements as possible during the design of the learning case; templates for these are provided by the authoring system It has been emphasized repeatedly that learners should be appropriately supported while working on a case (Cognition and Technology Group 1992; Gräsel 1997; Leutner 1992; Stark, Gruber et al 1998) Therefore, CASUS encourages the authors to make use of the following support elements: (1) Expert commentary The authors are asked to generate their comments on all important phases of a case to allude the learner to important or conflicting evidence and give reassuring feedback from a personal point of view (2) Mapping-Tool (Network-Tool) Every educational unit within a case is related to a graphical representation of the case: I allows the author to generate differential diagnostic hypothesis and connect them with the respective clinical findings Finally the authors are stimulated through tools within the software to develop a linear sequence of the flow of clinical information for the case before the media assembling process begins: they are assisted in choosing an appropriate (with a corresponding complexity) learning case An easy to use computer tool, the Befundmatrix (findings matrix), is made available for the collection and sequenzing of the case story (Figure 2) ZSfHD 3/99 With this tool the relevant clinical findings of a case are easily ordered by drag & drop interaction The next cucial step in the creation cycle of a learning case relates to the differential-diagnostic network Through this step the case author reconstructs the entire diagnostic reasoning process by connecting the clinical findings with related differential diagnostic hypothesis One emphasis of research on medical education is, how university education in the subject Medicine can be improved through increased implementation of problembased learning The expectation from problem-based learning is, on the one hand, that knowledge acquired in a course of studies can be better used in clinical practice, and, on the other hand, that the capability and motivation for lifelong learning is promoted Many theoretical and empirical studies concentrate on the development and evaluation of problem-based curricula for universities (e.g ACME-TRI Report, 1993; Albanese & Mitchell, 1993; Das Arztbild der Zukunft, 1993) The fundamental idea of problembased curricula centers around using clinical cases, which serve as the starting point for a study session of several hours in a small group led by a tutor Using these cases as examples the students analyze pathophysiological processes which characterize them, identify the course's educational objective, work on the appropriate material, and finally check to see if they have achieved their educational objective This kind of problem-based learning originated from Barrows (e.g Barrows, 1985; Barrows & Tamblyn, 1980) and was further developed and evaluated at various universities (e.g Armstrong, 1997; Gijslaers & Schmidt, 1990; Moore, Block, Style, & Mitchell, 1994; Tosteson, Adelstein & Carver, 1994; Williams, 1992) Another direction of research, which was pursued in the project being portrayed, concerned itself with the design of problem-based media that are both, implemented in university courses and offered in self-directed learning courses In particular computer-aided learning programs are suited as a medium for learning based on cases: The integration of multimedial elements makes a realistic and authentic presentation of clinical cases possible; through the use of interactive elements the students are encouraged to actively tackle the problems There are hardly any sound theoretical approaches for the design of problem-based learning programs — many of the existing programs were produced with intuitive didactic Furthermore, the development of learning cases with conventional authoring-systems requires information technology knowledge as well as a large investment of time — both have prevented instructors at universities and academic teaching hospitals from producing computer learning cases to any noticeable extent The goal of this project was to remedy this deficit: A learning system (CASUS) for medical education was developed comprising two parts (Figure 1): (1) CASUSLearning cases are multimedial learning programs for various clinical pictures which the students work through from the perspective of a physician (2) The CASUSAuthoring system should enable physicians to produce learning cases with minimal technical effort, and should support them in the didactical design of the case Figure A didactical concept that is oriented toward constructivist approaches to learning was developed for CASUS (M R G Fischer et al., 1996) These approaches have produced theoretical foundations, concrete instruction models, and empirical findings on problem-based learning environments (e.g Cognition and Technology Group at Vanderbilt, 1992, 1993; Collins, Brown & Newman, 1989; Duffy & Jonassen, 1992) In the following paper we will first introduce the problem-based concept of the authoring system CASUS We will then present results of a formative evaluation study of the authoring system In many development projects of learning software the formative evaluation has been given insufficient importance The learning programs are mostly produced under a deadline and difficulties and program flaws not become apparent until the finished programs are used — these are then seldom fixed after the fact In contrast, during the development of CASUS, a particular emphasis was placed on formative evaluation which ensured improvements in the program while it was being developed The questions to be posed in the formative evaluation study were determined by the objectives of CASUS: (1) Can clinically active authors produce learning cases on their own without technical difficulties? (2) Could the didactical concept of problem-based learning be put into action by authors designing their own cases? The Authoring System CASUS In the following section we will present the central principles of problem-based learning based on constructivist approaches, and scetch their realization in CASUS Constructivist approaches to learning and teaching emphasize two characteristics of learning environments which should impart useable knowledge (e.g Cognition and Technology Group at Vanderbilt, 1992; Collins et al., 1989): (1) Authenticity: The learning problems used should be similar to real problems in their central characteristics (2) Complexity: The complexity of the problems should not be reduced too much, and above all should be presented embedded in an authentic context In order to take these criteria into account, the structure of a learning case in CASUS is generally oriented toward the order of events in a genuine examination The steps in the examination make up the 'chapters' or didactic units of a learning case which a student works through one by one During the production of a learning case this scheme can be varied by adding further chapters or through 'jumps' or 'loops' In ZSfHD 3/99 ZSfHD 3/99 order to achieve the highest level of authenticity it is further requested of the authors that they include diverse multimedial elements (e.g audio, pictures, and video) while designing their cases A further principle is that knowledge is always actively constructed by the students in a given context From a constructivist perspective the learning environments should stimulate the students to actively elaborate and interpret the information in the learning material, and furthermore to use their prior knowledge on the problem, find their own way to a solution, and independently fall back on additional information sources In this way, the teachers are not serving up a prepackaged 'knowledge mix', rather they are supporting and correcting the independent learning process Correspondingly, the authors are encouraged by CASUS to use as many interactive elements as possible during the design of the learning case; these are provided by the authoring system consideration in CASUS: The students have the opportunity to compare their networks with an expert solution after every didactical unit Instructional supports for the Author The program CASUS is supposed to offer a sound instructional support for the design of cases To achieve this the authors are given a handbook which contains the basic ideas of the problem-based approach as they have been summarized above The principles of problem-based learning are furthermore demonstrated through an example case which the authors can view Finally the authors are stimulated through various tools to develop a comprehensive and well thought out concept for a learning case before the concrete programming of the case begins For example, they are assisted in choosing an appropriate (with a corresponding complexity) learning case An easy to use computer tool, the Befundmatrix (findings matrix), is made available to the authors for the collection and ordering of patient data (Figure 3) Authentic and complex problems should stimulate the students to active and constructive learning processes However, studies in Medicine show that students are often overwhelmed by the complexity of the problems and are not stimulated to appropriate learning processes (Gräsel & Mandl, 1993; Gräsel 1997) Similar results have also been found in other domains (Cognition and Technology Group, 1992; Leutner, 1992; Stark, Gruber, Renkl & Mandl, 1998) A conclusion of these investigations is that students should be appropriately supported while working on a case Therefore, CASUS encourages the authors to take the following support elements into account: (1) Expert commentary The authors are asked to prepare comments, which contain an expert's views on the case These comments should demonstrate the process of diagnosing to the students during the entire time they are working on the case, and thus stimulate them to form their own hypotheses In constructivist approaches the context-related articulations of an expert play a central role (e.g Collins et al., 1989) For learning with computer-aided cases in Medicine, empirical investigations showed that comprehensive explanations by experts contribute to the amount students learn while working on a case (Gräsel & Mandl, 1993; Gräsel, 1997) (2) Mapping-Tool (Network-Tool) Every didactic section ends for the students with the request to graphically represent the findings of the case, their interrelations and their connections to hypotheses (Figure 2) Findings show that graphical representation of a case with a mapping process is an appropriate aid for learning with computer-presented cases in Medicine (F Fischer, 1998; F Fischer, Gräsel, Kittel & Mandl, 1996) For one thing, the use of a mapping process supports the application of suitable strategies of diagnosis in working on a case For another, the students themselves can evaluate at a glance the completeness and coherence of their own problem solution, and correct themselves accordingly In an empirical study it was shown that a student's reflection on his/her solution is particularly effective for learning when the student is encouraged to compare his/her solution with that of an expert (F Fischer et al., 1996) This finding was taken into With this tool the central findings of a case are ordered with drag & drop interaction helping to plan the rough outline of the entire learning case Furthermore, in this graphic organization of the learning case it is already possible to mark positions where multimedial material will be necessary The next step in the programming of a learning case consists of producing the differential-diagnostic network for every single program step Through this step the author must reconstruct the entire process of his reaching a diagnosis; this results not only in a central support for the students, but also in a general idea of the goals and content of his learning case Only after this groundwork does the actual design of the learning case begin, namely the work with the individual screens The authors are asked to design the screens one by one as chapters In order to arrange the information for each page there are seven different types of forms on 'index' cards (Figure 4) available to the authors (for example: card type combines information, multimedia, and question-answer frames) components in respect to their practicability and understandability while the program was still in the early stages of development The study which is reported here was aimed mainly at improving the authoring program CASUS For this purpose, detailed information about difficulties in producing a case was necessary; thus, a qualitative research method was used: The process of producing a case was observed and analyzed in order to derive consequences for further development from the results Research Question of the study The study pursued mainly the question of which difficulties arise while producing a case with CASUS and which measures are necessary in order to remedy these This was done by investigating both the technical aspects (use of the technical functions of the program) as well as the didactical aspects (application of the concept of problembased learning) of producing a case Additionally, the acceptance and motivation of the authors in the use of the authoring system were elements to be investigated Method After the completion of the first beta-version of CASUS four persons were asked to produce a learning case with the authoring system in their respective specialty areas The test subjects came from the target group of the authoring system, namely they are all experienced clinicians who teach at the university level and who are open to the idea of problem-based learning The authors had neither any special prior technical experience (i.e with other authoring systems) nor special prior knowledge about the concept of problem-based learning Before the subjects began with the programming of a case they were told (1) to work through the introductory chapter of the handbook, (2) to take a look at a demonstration case, (3) to choose a case which they considered suitable for their students and (4) to determine and produce the necessary multimedia material The conditions under which a case was produced were to be as similar as possible to those which later authors would experience Therefore, the authors worked mostly independently with CASUS The following data were recorded: Acceptance of the authoring system and motivation in case producing In a structured interview after the production of a case, and with the help of a questionnaire, the acceptance of the authoring system and the motivation while producing a case were recorded This paper will report on the primary evaluation study which was carried out with the first beta-version of CASUS Individual components of the authoring system (i.e the handbook, the Befundmatrix (findings matrix)) had already been evaluated before this study Experts in Medicine and Instructional Research evaluated the components in the sense of a quality analysis Furthermore, potential authors assessed these ZSfHD 3/99 Difficulties and interventions in case producing The process of case producing was observed by a member of the CASUS development-team who wrote a detailed protocol Whenever difficulties with case producing arose the observer intervened in the process Based on a question checklist he asked the author about the causes of the problems and possiblities for remedying them Thus, the observer and the author worked out a solution to the problem area together The computer logfile protocols as ZSfHD 3/99 well as a subsequent interview were further sources of information about problems and their solution in the interaction with CASUS Both technical and didactical difficulties were recorded by the observer The observers' protocols and the subsequent interviews were evaluated by structured extraction of key information The difficulties were classified as being either technical or didactical In the following, only difficulties of interaction with the program are reported; software problems (bugs) are not taken into consideration Of course, these were recorded and fixed in further program development, too Results Acceptance and Motivation The authoring system was met with a high level of acceptance: For one, the multiple-choice questions yielded — almost without exception — high acceptance values; for another, positive assessments of the learning system predominated in the interviews (e.g "the authoring system is altogether well suited for producing learning cases" or "I would also recommend the authoring system to others") The critical comments referred mostly to software problems (e.g computer crashes) which were still present in the first beta- version The picture painted in the accounts about motivation was more differentiated: Though it was expressed in both, the questionnaires and the interviews, that the authors' intrinsic motivation was quite high (for example they expressed that the interaction with the program was fun), it became clear in three of the interviews that the authors would also appreciate an extrinsic motivation to produce a case — for example the possibility to publish learning cases At the very least, it is seen as necessary to get time off from the clinical work for the duration of the case production Difficulties with the technical interaction and necessary supports The quality of the user interface and the usability of the program were assessed by the subjects for the most part positively In two interviews the drag and drop technique in particular were described as very intuitive and easy to learn This positive assessment is in agreement with the analysis of the case production: Difficulties in the utilization of the program barely arose while producing a case with CASUS Accordingly, there were few places where the observers had to intervene for the purpose of giving technical support Also in the subsequent interviews, the suggestions of changes and supports in a technical sense were few and easy to realize (e.g easier printing of screens) Difficulties with the realization of the didactic concept and necessary supports The application of problem-based learning as an instructional concept in the production of a case turned out to be much more problematic The handbook was considered very comprehensible by all the authors, and they believed also that with the help of the handbook they had understood the basic idea of problem-based learning; however, during the production of the case many questions and difficulties arose concerning preparing the medical content in the design of a case One main difficulty which all four subjects experienced throughout the production of a case was in dealing with the learning objectives For one thing it was troublesome for the authors to formulate learning objectives based on the case For another, they often only succeeded with coaching from the observer in designing a didactic unit or screen from the viewpoint of a specific learning objective Two subjects also required significant support with the design of differential-diagnostic networks; above all, the case authors had the difficulty here of choosing relevant findings and suspected diagnoses appropriate to the target group Furthermore, three of the subjects were barely capable of choosing and designing appropriate interactive elements without support (e.g question and answer sequences, cards with multimedial components) An analysis of the points where intervention was necessary for the instructional design can be summed up as follows: Almost all interventions for instructional design were directly related to the content of the specific case Standardized instructional support for the authors — whether in a handbook or as online help — would have been of use with very few of the difficulties The problems the authors had were not instructional, but rather a mixture between the specific content of the case and questions how to transfer these content into a case-based learning program Concept-mappping techniques support the hypothetical-deductive problem solving process In medicine through the use of these techniques an improvement in both quantity and quality of diagnostic hypotheses can be demonstrated The case-based learning system CASUS provides a mapping-tool for the visualization of medical differential diagnoses, which employs drag&drop interactivity to connect the critical findings of a learning case to the respective hypothesis by weighed links We evaluated the use of the CASUS-mapping component in internal medicine cases in the rd clinical semester at the University of Munich: 439 case sessions were recorded yielding 265 differential diagnostic networks with at least two hypothesis (60.3%) This paper describes the characteristics of the student differentials and compares them with the solutions derived from experts Further it provides perspectives for the use of this data in future feedback processes Keywords: Mapping-Tool, Case-based Learning, Differential Diagnosis, Hypothetical Deductive Process, CASUS-Lernsystem Einleitung: Der Einsatz von Concept-Mapping-Techniken zur Unterstützung des Methoden: Problemlöseprozesses ist in verschiedenen Domänen etabliert [1] Fischer et al Das Mapping-Werkzeug zur Erstellung differentialdiagnostischer Netzwerke wurde konnten zur 1996 mit der Object Modelling Technic von J Rumbaugh entworfen Die Differentialdiagnose von Anämien [2] zeigen, das ein einfach zu bedienendes Implementierung erfolgte mit der Entwicklungsumgebung Metrowerks CodeWarrior® Mapping-Werkzeug zu einer signifikanten Zunahme der gebildeten diagnostischen in der Programmiersprache C++ Die Daten werden in einer relationalen Datenbank Hypothesen und der Verknüpfungen der Hypothesen mit klinischen Befunden bei (ORACLE 8.0.5) abgelegt [4] Die dieser Arbeit zu Grunde liegenden Auswertungen Medizinstudenten im klinischen Studienabschnitt führt: Dabei profitierten diejenigen erfolgten über SQL-Anweisungen mit dem graphischen Datenbank-Explorer Voyant® für Studenten die mit Medizin dem anhand höchsten eines Vorwissen fallbasierten am Lernsystems meisten von dieser Visualisierungsmöglichkeit Darüberhinaus zeigte sich ein nicht-signifikanter Trend zu mehr korrekten Hypothesen bei der Studentengruppe, die das Mapping-Werkzeug benutzte Das CASUS-Lernsystem [3] erlaubt die rasche Entwicklung von fallbasierten Lerneinheiten für Studierende der Medizin Die Fallautoren benötigen keine Programmierkenntnisse und werden didaktisch bei der Fallstrukturierung unterstützt Die Fälle sind linear strukturiert und werden von den Studierenden Bildschirmkarte für Bildschirmkarte bearbeitet Bei der Bearbeitung müssen klinische Informationen (sog Befunde) gesammelt und in Frage-Antwort Dialogen interpretiert werden Ein Fall gliedert sich in didaktische Einheiten, für die jeweils explizite Lernziele von den Fallautoren formuliert werden Am Ende jeder didaktischen Einheit eines Falles wird ein Mapping-Werkzeug aufgerufen, mit dem die Befunde per Drag&Drop-Interaktivität mit diagnostischen Hypothesen in Beziehung gesetzt werden sollen Der Studierende kann das eigene differentialdiagnostische Netzwerk mit dem des klinischen Experten vergleichen In dieser Arbeit wird die Evaluation der von Studierenden im klinischen Studienabschnitt erstellten differentialdiagnostischen Netzwerke dargestellt Welche Bedeutung hat eine tutorielle Betreuung für die Benutzung des Mapping-Werkzeugs? Welche weiteren Entwicklungs- und Evaluationsschritte sind erforderlich, um solche Werkzeuge im Feedbackprozess zwischen Lehrenden und Lernenden nutzbringend einzusetzen? ZSfHD 3/99 ZSfHD 3/99 Abb. 1: Expertennetzwerk des CASUS­Lernfalles zum Thema "Obere gastrointestinale Blutung": Die blauen Elemente repräsentieren klinische Befunde, die diagnostischen Hypothesen sind rot und Therapien grün dargestellt In Abbildung ist exemplarisch das Netzwerk eines Experten dargestellt Die blau eingefärbten Befundelemente werden der zeitlichen Abfolge des Lernfalles entsprechend Schritt für Schritt per Drag&Drop auf die Arbeitsfläche positioniert Dabei läßt sich des differentialdiagnostische Netzwerk jederzeit "zurückblättern" (Pfeile unterhalb der Arbeitsfläche in Abbildung 1) In rot dargestellte Hypothesen werden ebenfalls per Drag&Drop neu erstellt und mit gewichteten Verbindungen zu den entsprechenden Befunden versehen Die blauen Linien stellen dabei bestätigende und die roten Linien widerlegende Verknüpfungen dar Es stehen jeweils drei Verbindungsstufen zur Verfügung Abb. 2: Dialogfenster bei der Erstellung einer neuen Zuordnung zwischen klinischem Befund und Hypothese: Positive Verknüpfungen sind in drei Stufen in blau, negative in rot auswählbar. Jede Verknüpfung sollte vom Fallautor und dem Studierenden begründet werden. Wenn verfügbar, lassen sich Daten zu Prävalenz, Sensitivität und Spezifität eines Befundes im Bezug auf die entsprechende Hypothese eingeben Tabelle 1: Quantitative Auswertung der differentialdiagnostischen Netzwerke der Studierenden; Fall 1 mit tutorieller Unterstützung, die Fälle 2 bis 5 im Selbststudium Die statistische Analyse der Nutzerdaten zur Erstellung der differentialdiagnostischen Netzwerke wurde mit dem Chi-Quadrat Test in StatView®durchgeführt Ergebnisse: Die Daten zeigen, daß die Anzahl der Fallbearbeitungen offensichtlich stark von der Relevanz der Fälle für die Anrechnung eines Testates beeinflußt wurde Die Wir evaluierten Studierende aus dem klinischen Semester der Ludwig-Maximilians- Anzahl auswertbarer Netzwerke wurde wesentlich von der Anwesenheit eines Tutors Universität München Zu sechs CASUS-Lernfällen aus der Inneren Medizin konnten beeinflußt, der die Handhabung des Werkzeuges erklärte Im Selbststudium bei den wir 439 Lernsitzungen aufzeichnen Die Fallbearbeitung lag zwischen 60 und 90 Fällen bis erstellten signifikant weniger Studierende ein Netzwerk (p < 0,01) Die Minuten pro Lernfall Die Lernfälle wurden vorlesungsbegleitend und abgestimmt auf Anzahl der generierten Hypothesen unterschied sich dagegen im Vergleich der die Vorlesungsthemen angeboten und sequentiell alle 10 bis 14 Tage im Semester einzelnen Fälle nicht signifikant voneinander Für die Anzahl der Verknüpfungen freigeschaltet Dabei wiesen 265 Logfiles zu den einzelnen Fallbearbeitungen zwischen Befunden und Hypothesen und die Anzahl der Verknüpfungen pro differentialdiagnostische Netzwerke (60,3%) mit mehr als Hypothesen auf und Hypothese ergab sich nur für den sechsten Fall ein signifikanter Unterschied (p < wurden als relevant angesehen und ausgewertet Fall (siehe Tabelle 1) wies den 0,05) Dieser Fall wurde allerdings nur von zwölf Studenten bearbeitet höchsten Anteil auswertbarer Daten auf, weil dieser Fall unter tutorieller Anleitung Möglicherweise handelt es sich um eine hochmotivierte Untergruppe von Studenten, bearbeitet und als Testat für die Vorlesung Innere Medizin angerechnet wurde Zwei die nicht repräsentativ sind Dafür spricht die Tatsache, daß sie am Semesterende vor weitere frei wählbare Fälle qualifizierten die Studierenden für ein weiteres Testat Die den Abschlußklausuren einen Fall ohne Testatrelevanz bearbeiteten Fälle bis wurden im Selbststudium ohne Anleitung bearbeitet allnummer Gesamt allbearbeitungen 119 103 63 102 40 12 439 nzahl auswertbarer etzwerke (%) 109 51 25 59 16 265 (91,6) (49,5) (39,7) (57,8) (40) (41,7) (60,3) ypothesen pro Fall 2,8 3,2 2,5 2,8 2,3 3,0 2,82 erknüpfungen efund-Hypothese 9,4 8,7 5,8 9,2 5,3 14,6 8,69 3,4 2,7 2,3 3,3 2,3 4,9 3,08 erknüpfungen ypothese Zusammenfassung: Das Mapping-Werkzeug Hypothesenbildung im zur Unterstützung CASUS-Lernsystem der differentialdiagnostischen weist unter tutoriellen Betreuungsbedingungen eine hohe Akzeptanz auf Im Selbststudium erstellt dagegen nur ein Teil der Studierenden ein Netzwerk Die Qualität der Netzwerke unterscheidet sich unter beiden Bedingungen nicht Die Netzwerkdaten der klinischen Experten wurden bereits MeSH-codiert und pro sind somit inhaltlich eindeutig zuzuordnen In einem nächsten Evaluationsschritt werden die Netzwerkdaten der Studierenden ebenfalls qualitativ mithilfe eines halbautomatischen MeSH-Codiersystems [5] analysiert und mit den Daten der ZSfHD 3/99 ZSfHD 3/99 Experten verglichen Diese Vergleichsdaten sollen dann in die Feedbackkomponente des CASUS-Lernsystems integriert werden Ein Dozent kann die diagnostischen Unstimmigkeiten zwischen dem Vorgehen des Experten und dem der Studierenden als Basis für eine Diskussion mit den Studierenden verwenden Ob das differentialdiagnostische Netzwerk des Experten den Studierenden erst am Ende eines Falles gezeigt werden sollte und welchen Einfluß dies auf die Quantität und Qualität der Netzwerke hat, wird eine weitere Evaluation zeigen Discussion With the authoring system CASUS physicians can produce learning cases even without any special prior knowledge with regard to technical aspects The technical difficulties which arose during the formative evaluation of the authoring system could be taken into consideration without too much trouble when the program was revised The handbook and the online help appear — based on these results — to be sufficient to use CASUS for the production of cases with little technical effort Altogether, it can be expected that authors who have used other computer applications (and who otherwise have no great computer knowledge) can learn the technical operation of the program quickly and without difficulty In the didactical respect, there were far more problems with the production of cases Only a few of the difficulties encountered can be addressed in the handbook or online help since the problems were very specific to the content of individual cases This leads to the conclusion that it is at least helpful if not necessary to provide the authors with support in the didactical aspect of the content At the very least, this is the case with authors who not yet have comprehensive experience in problem-based teaching This result is in agreement with findings on the introduction of problembased curricula: In evaluation studies it was shown that instructors felt themselves to be not up to the task of teaching a problem-based course, and they desired more support (Albanese & Mitchell, 1993) Because of the experiences in the formative evaluation study, it seems that an appropriate measure for the CASUS authors would be a tutor model Close cooperation between the author and a person who demonstrates expertise in both a professional and instructional sense appears to be a good way to ensure the quality of the learning cases The development of a guide for tutoring of authors is only one of the goals which will be pursued within the scope of this project in the near future The emphasis will now be placed on using the authoring system CASUS to produce suitable learning cases for different content-areas and levels of study This is connected with investigations into the effectiveness of learning with CASUS cases Thus far, an initial evaluation with a prototypical case was performed only to determine the acceptance and learning motivation among students It was shown that the students generally worked on the case with a high level of motivation and most expressed the wish that cases of this type would be better integrated in the course of studies This judgement of the learning cases proved to be independent of the technical competence of the students In further investigations the question will be pursued on the basis of constructivist approaches, to what extent computer-based learning cases are suitable for imparting coherent and application-oriented knowledge An emphasis will be placed on finding differentiated quality characteristics of learning cases on a theoretical and empirical basis With these subsequent empirical studies and the efforts to implement the existing cases at different universities, it will be attempted to contribute to the improvement of medical education An important field of application is the supplementation of clinical courses with key symptoms and diseases in the case format As an integral part of a course, the cases will ensure the even exposure of students to these problems when there is no real patient with the respective problem available References series as a case study In T M Duffy, J Lowyck, D H Jonassen & T M Welsh (Eds.), Designing environments for constructive learning (pp 9-36) Berlin: Springer ACME-TRI Report (1993) Educating medical students Collins, A., Brown, J S & Newman, S E (1989) Cognitive Assessing change in medical education - the road to apprenticeship: Teaching the crafts of reading, writing, and implementation Academic Medicine, 68 (6, Supplement), 1-48 mathematics In L B Resnick (Ed.), Knowing, learning, and Albanese, M A & Mitchell, S (1993) Problem-based learning: A review of literature on its outcomes and implementation issues Academic Medicine, 68, 52-81 Armstrong, E G (1997) A Hybrid Model of Problem-based instruction Essays in the honour of Robert Glaser (pp 453-494) Hillsdale: Erlbaum Das Arztbild der Zukunft (1993) Analysen künftiger Anforderungen an den Arzt - Konsequenzen für die Ausbildung und Learning In D Bound & G Feletti (Eds.) The Challenge of Wege zu ihrer Reform [Analysis of future standards for physicians Problem-Based Learning (pp 137-150) London: Kogan Page Consequences for the reform of medical education] Gerlingen: Barrows, H S (1985) How to design a problem-based curriculum for the preclinical years New York: Springer Barrows, H S & Tamblyn, R (1980) Problem-based learning New York: Springer Cognition and Technology Group at Vanderbilt (1992) The Jasper series as an example of anchored instruction: Theory, Blecher Duffy, T M & Jonassen, D H (1992) Constructivism: New implications for instructional technology In T M Duffy & D H Jonassen (Eds.), Constructivism and the technology of instruction: A conversation (pp 1-16) Hillsdale: Erlbaum Fischer, F (1998) Mappingverfahren als Werkzeug für das program, description, and assessment data Educational problemorientierte Lernen Entwicklung und empirische Psychologist, 27, 291-315 Untersuchung eines computerbasierten Mappingverfahrens in der Cognition and Technology Group at Vanderbilt (1993) Domäne Medizin [Concept mapping as a tool for problem-oriented Designing learning environments that support thinking: The Jasper ZSfHD 3/99 ZSfHD 3/99 learning Development and evaluation of a computer-based mapping oriented learning Strategy use and possibilities for design] technique] Frankfurt: Lang Göttingen: Hogrefe Fischer, F., Gräsel, C., Kittel, A & Mandl, H (1996) Gräsel, C & Mandl, H (1993) Förderung des Erwerbs Entwicklung und Untersuchung eines computerbasierten diagnostischer Strategien in fallbasierten Lernumgebungen Mappingverfahrens zur Strukturierung komplexer Information [A [Fostering the acquisition of diagnostic strategies in case-based computer-based mapping technique for structuring complex learning environments] Unterrichtswissenschaft, 21, 355-370 information: Development and some preliminary data] Psychologie in Erziehung und Unterricht, 43, 266-280 Fischer, M R G., Schauer, S., Gräsel, C., Baehring, T., Leutner, D (1992) Adaptive Lehrsysteme [Adaptive systems for learning] Weinheim: Psychologie Verlags Union Moore, G T., Block, S D., Style, C B & Mitchell, R Mandl, H., Gärtner, R., Scherbaum, W & Scriba, P C (1996) (1994) The influence of the New Pathway curriculum on Harvard Modellversuch CASUS - ein computergestütztes Autorensystem für Medical students Academic Medicine, 69, 983-989 die problemorientierte Lehre in der Medizin [CASUS - a computer- Stark, R., Gruber, H., Renkl, A & Mandl, H (1998) based authoring system for problem-based learning in Medicine] Instructional effects in complex learning: Do objective and Zeitschrift für ärtzliche Fortbildung, 90, 385-389 subjective learning outcomes converge? Learning and Instruction, Gijselaers, W H & Schmidt, H G (1990) Development and evaluation of a causal model of problem-based learning In A 8, 117-129 Tosteson, D C., Adelstein, S J & Carver, S T (Eds) M Nooman, H G Schmidt & E S Ezzat (Eds.), Innovation in (1994) New Pathways to Medical Education: Learning to Learn at medical education: An evaluation of its present status (pp 95-113) Harvard Medical School Cambridge: Harvard University Press New York: Springer Gräsel, C (1997) Problemorientiertes Lernen Strategieanwendung und Gestaltungsmöglichkeiten [Problem- Williams, S M (1992) Putting case-based instruction into context: Examples from legal and medical education The Journal of the Learning Sciences, 2, 367-427 1 Fischer F: Mappingverfahren als kognitive Werkzeuge für problemorientiertes Lernen Europäischer Verlag der Wissenschaften (Frankfurt am Main) 1998 Fischer F, Gräsel C, Kittel A & Mandl H: Entwicklung und Untersuchung eines computerbasierten Mappingverfahrens zur Strukturierung komplexer Information Psychologie in Erziehung und Unterricht 1996; 43:266-280 3.  Fischer MR, Schauer S, Gräsel C, et al.: Modellversuch CASUS Ein computergestütztes Autorensystem für die Problemorientierte Lehre in der Medizin Zeitschrift für Ärztliche Fortbildung 1996; 90:385-9 Adler M: Entwicklung einer differentialdiagnostischen Netzwerk-Komponente für die medizinische multimediale Lerndatenbank CASUS Diplomarbeit am Institut für Informatik der Technischen Universität München 1996 Holzer M F, Dietrich JW & Fischer MR: Freitextverarbeitung und Verschlagwortung in medizinischen Lernprogrammen: Ein Vergleich von SNOMED und MeSH In: Methoden der Medizinischen Informatik, Biometrie und Epidemiologie in der modernen Informationsgesellschaft E Greiser und M Wischnewsky (Hrsg) 1998 München, MMV: 282-6 ZSfHD 3/99 ZSfHD 3/99 ... clinical practice, and, on the other hand, that the capability and motivation for lifelong learning is promoted Many theoretical and empirical studies concentrate on the development and evaluation... use of these techniques an improvement in both quantity and quality of diagnostic hypotheses can be demonstrated The case-based learning system CASUS provides a mapping -tool for the visualization... Constructivist approaches to learning and teaching emphasize two characteristics of learning environments which should impart useable knowledge (e.g Cognition and Technology Group at Vanderbilt, 1992; Collins