Linköpings universitet
Linköping University | Department of Computer Science
Bachelor thesis, 18 ECTS | Cognitive Science
Spring 2017 | LIU-IDA/KOGVET-G–17/012—SE
Saving Lives More Efficiently:
A First Step in Designing a
Visual Language
–
Creating and evaluating a visual language aimed at the
medical field
Rädda liv på ett mer effektivt sätt: skapande och
evaluerande av ett visuellt språk inom vården
Jasmina Jahic
Supervisor : Arne Jönsson Examiner : Henrik Danielsson
Upphovsrätt
Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare – under 25 år från publiceringsdatum under förutsättning att inga extraordinära omständigheter uppstår. Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner, skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för ickekommersiell forskning och för undervisning. Överföring av upphovsrätten vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av dokumentet kräver upphovsmannens medgivande. För att garantera äktheten, säkerheten och tillgängligheten finns lösningar av teknisk och admin-istrativ art. Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i den omfattning som god sed kräver vid användning av dokumentet på ovan beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan form eller i sådant sam-manhang som är kränkande för upphovsmannenslitterära eller konstnärliga anseende eller egenart. För ytterligare information om Linköping University Electronic Press se förlagets hemsida http://www.ep.liu.se/.
Copyright
The publishers will keep this document online on the Internet – or its possible replacement – for a period of 25 years starting from the date of publication barring exceptional circum-stances. The online availability of the document implies permanent permission for anyone to read, to download, or to print out single copies for his/hers own use and to use it unchanged for non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional upon the con-sent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security and accessibility. According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement. For additional information about the Linköping Uni-versity Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its www home page: http://www.ep.liu.se/.
c
Abstract
The aim of this study is to create and evaluate a grammar for a visual language, which is to be used in the medical domain. The grammar was created with gestalt laws in mind. The grammar must also fit, in part, with the ontology of the clinical terminology SNOMED CT. Evaluation was conducted with think aloud tests and with complementary semi-structured interviews. The participants overall believed that a complementary visual language for pa-tient journals would be useful, and the majority understood the logic behind the grammar. Most complaints were about some of the pictograms in the icons of the grammar, which were unclear, and some requested a clear time axis to be included. Another suggestion was to add pop-up boxes with more detailed information. This is only the first step in a big project and there is potential for many improvements - future development of this visual language should be done in iterations by teams of complementing competencies.
Acknowledgments
To Arne, for being my supervisor and giving me guidance, as well as for making sure that I had fun during the project.
To Daniel, Erik and Mikael, for answering my many questions and being patient with me.
To Elin, Isabella, Johanna and Vida, for the support and encouragement. To Griffin, Justin and Travis McElroy for creating quality content and always providing a laugh when needed the most.
Contents
Abstract iii
Acknowledgments iv
Contents v
List of Figures vii
List of Tables viii
1 Introduction 1
1.1 Motivation . . . 1
1.2 Aim . . . 1
1.3 Delimitations . . . 2
2 Theory 3 2.1 The creation of a grammar . . . 3
2.1.1 Semiotics . . . 3
2.1.2 Gestalt laws . . . 4
2.2 Methodology . . . 6
2.2.1 Think aloud test . . . 6
2.2.2 Semi-structured interviews . . . 6
2.3 Related works . . . 7
2.3.1 SNOMED CT . . . 7
2.3.2 Visualisation of Concepts in Medicine, VCM . . . 7
3 Method 10 3.1 Designing the grammar and icons . . . 10
3.2 Participants . . . 10 3.3 Test structure . . . 11 3.4 Test procedure . . . 13 4 Results 14 4.1 The grammar . . . 14 4.2 The interviews . . . 16 5 Discussion 18 5.1 Results . . . 18 5.2 Method . . . 20 6 Conclusion 22 References 24
A Icons 26
B Test Instructions 28
List of Figures
2.1 The principle of proximity. Adapted from “Laws of Organization in Perceptual Forms” by M. Wertheimer in A Source Book of Gestalt Psychology edited by W. D.
Ellis, 1938, p. 71-88. Copyright 1938 by Gestalt Legacy Press. . . 4
2.2 The principle of similarity. Adapted from “Laws of Organization in Perceptual Forms” by M. Wertheimer in A Source Book of Gestalt Psychology edited by W. D. Ellis, 1938, p. 71-88. Copyright 1938 by Gestalt Legacy Press. . . 4
2.3 The principle of continuity. Adapted from “Laws of Organization in Perceptual Forms” by M. Wertheimer in A Source Book of Gestalt Psychology edited by W. D. Ellis, 1938, p. 71-88. Copyright 1938 by Gestalt Legacy Press. . . 5
2.4 The principle of figure/ground, illustrated with the Rubin vase. . . 5
3.1 The first page of the test with the grammar and icons. . . 12
3.2 The second page of the test with the grammar and icons. . . 12
4.1 The basic shapes and colours for the grammar. . . 14
4.2 Up- and down-arrows. . . 14
4.3 Left: Compounded icons to represent high blood pressure. Right: One icon that represents hypercholesterolemia. . . 15
4.4 Example of the placement of icons. Grouping by proximity and similarity by icon shape and colour. . . 15
List of Tables
3.1 Participant demographics. . . 11 4.1 Category mapping. . . 14
Chapter 1
Introduction
The beginning of this section explains why it is worthwhile to explore visual representation of information and why this study does so, followed by the more specific aim and purpose of this study. Following this is a short summary of what will not be explored in the study.
There are many ways to represent information. Arguably, the most dominant form is through text. However, the more text that a person has to read through to find specific infor-mation, the more mental resources are needed. Patient journals and other kinds of medical documentation are no exception. Doctors, nurses and other medical staff need to keep track of many patients within a limited amount of time. A common way to lessen the burden of reading several pages of text is to complement it with visual information, whether it be pho-tographs, illustrations, symbols or icons. A benefit of using visual information is that human beings can process more information, as long as they are good representations of the concepts behind the pictures - in other words, it is easer to recognise visual information. In that case, it would not be so strange to implement visual information into different types of medical documents.
1.1
Motivation
Region Östergötland, along with many other healthcare institutions in Sweden, are working on digitalising their documents and making them as accessible as possible. To do this they would like to implement SNOMED CT, and together with it integrate icons and symbols into clinical documents that will make the information easier to comprehend. This study is a first step in realising that vision.
1.2
Aim
The aim of this study is to explore how a grammar for a visual language in the medical field may look like.
The research questions are:
1. How can a grammar for a visual language look like when it is partly based on the ontology of SNOMED CT?
2. What level of detail of a visual language is appropriate for people in the medical field when reading patient journals?
3. Which information do people in the medical field need in order to understand a medical visual language?
1.3. Delimitations
1.3
Delimitations
The grammar that has been created has been done so with strictly the medical domain in mind. It may be possible to apply the grammar on other domains, however this has not been tested.
Icons were designed to visualise the grammar, but not much research has been done on how the pictograms in the icons should look like to be intuitive and informative. The main focus and background of the elements of the icons lie in cognitive psychology, e.g. which shapes to use.
This study did not delve into whether visual information is better than purely textual, since earlier studies have already proved that humans can more rapidly recognise graphical and visual information than pure text (Nelson, Reed, & Walling, 1976; Wiedenbeck, 1999).
Chapter 2
Theory
This chapter begins by bringing up some components needed to have in mind when creating a grammar. After this, the methods used for evaluation are explained. The chapter is closed after introducing related works.
2.1
The creation of a grammar
Grammar for each and any language will always be complex, whether it be a natural or a constructed one. A natural language is one that has evolved “naturally” throughout history, such as English. On the other hand, a constructed language is one that is artificially created, such as Esperanto and the Elvish languages made by J. R. R. Tolkien. The latter ones are fairly matured languages, whereas this study will be the first step in creating something akin to them. Even if it is a visual language and its grammar will not need to have the many complex elements that are in a written and spoken one, the very foundation will have its similarities.
The next sections describe the tools that were used in creating a grammar for a visual lan-guage. First, we get into meaning-making and representations of concepts, namely semiotics. After this some of the gestalt laws are described.
2.1.1
Semiotics
Semiotics is the study of signs and meaning-making. Here, a sign is defined as “something that stands for something else”. This means that everything in our environment that repre-sents something else, whether it be language, images or even clothing, is a sign. This also means that culture has a heavy role in the field of semiotics (Danesi, 2007). A good exam-ple of cultural influence on symbols is different kinds of hand signs - the sign for “OK” in America means “money” in Japan, “zero” in France and is seen as offensive in Russia.
This example can also be used to explain the triad of Charles Sanders Peirce, one of the founding fathers of modern semiotics. Danesi (2007) writes that Peirce talked about the rep-resentamen (the form/actual sign; “something that represents something else”), the object (the concept; “that which is to be represented”) and the interpretant (the meaning; “that which changes depending on culture and context”). Peirce also wrote about three different kinds of signs, namely icons, indices and symbols. The differences are that an icon is very literal in what it is representing. Even if one may not be familiar of the culture or context, one can understand the icon if someone explains how it resembles the concept it is representing. Indexicality has to do with spatiality and relating objects to each other. An example is of someone pointing at something and using words such as “here” or “there” depending on how far away an object is in relation to the pointer’s own position. Symbols are signs that are learned and used by convention - an example that Danesi uses is that of the Christian cross. At the beginning it represented the cross on which Christ was crucified, but through the times it has instead become associated with the religion of Christianity. In this study of
2.1. The creation of a grammar
creating a grammar for a visual language, “icon” and “symbol” follow these definitions of Peirce and are not interchangeable.
2.1.2
Gestalt laws
Gestalt laws are principles in psychology, popularised by Max Wertheimer, Kurt Koffka and Wolfgang Köhler. The principles offer an explanation as to how humans perceive objects. The overarching law for all of the gestalt principles is the law of Prägnanz (often translated as “Good Figure” or law of simplicity). This law states that humans tend to, as simply as possible, perceive, organise and unify visual arrays into stable and coherent forms (seeing it as a “whole”)(Sternberg & Sternberg, 2011). There are many principles which belong to the gestalt laws, and only the ones being used in this study will be described here.
The principles used in this study are the principles of proximity, similarity, continuity and figure/ground. The principle of proximity states that humans group together elements that are spatially close together(Sternberg & Sternberg, 2011; Wertheimer, 1938). This is illustrated in Figure 2.1, where the left section looks like the dots are placed vertically, while in the right section it seems like they create a horizontal line. These “lines” really only are dots and nothing else - but because they are placed in the way they are, we perceive them as being “vertical” or “horizontal”.
Figure 2.1: The principle of proximity. Adapted from “Laws of Organization in Perceptual Forms” by M. Wertheimer in A Source Book of Gestalt Psychology edited by W. D. Ellis, 1938, p. 71-88. Copyright 1938 by Gestalt Legacy Press.
The principle of similarity builds on the basis that humans group together objects that are similar in some way. One example can be shown in Figure 2.2. Even if it is the same prox-imity between the dots and the circles, the human will perceive the dots as belonging to each other as one group, and the circles as another. According to Wertheimer (1938), these aspects can be controlled to some degree - the further apart the elements are, the more predominant the aspect of similarity becomes. Kolers (1969), in his article Some Formal Characteristics of Pictograms, also brings up how one may group and categorise information by using simi-larity. The example he brings up is that of traffic signs - information regarding “dangerous situations” may be indicated by triangular shapes, while “types of permitted traffic” may be indicated by circles.
Figure 2.2: The principle of similarity. Adapted from “Laws of Organization in Perceptual Forms” by M. Wertheimer in A Source Book of Gestalt Psychology edited by W. D. Ellis, 1938, p. 71-88. Copyright 1938 by Gestalt Legacy Press.
When an object is partly obscured by another object, the human will know that the ob-scured part is still there - it is not as if the obob-scured part suddenly stops existing. This can
2.1. The creation of a grammar
partly be explained by the principle of continuity. If a part of a table is obscured by a set of books, we do not assume that the table ends where one side of the books begin and begin again where the books end. We can follow the line of the edge of the table and “see” it be-hind the books. Another example can be seen in Figure 2.3. Here, two curved lines intersect. Usually a human would perceive the lines as AC and BD, rather than AD and BC (Sternberg & Sternberg, 2011; Wertheimer, 1938).
Figure 2.3: The principle of continuity. Adapted from “Laws of Organization in Perceptual Forms” by M. Wertheimer in A Source Book of Gestalt Psychology edited by W. D. Ellis, 1938, p. 71-88. Copyright 1938 by Gestalt Legacy Press.
Kurt Koffka is known to have said, “The whole is different from the sum of its parts” (Hei-der, 1977). What he probably meant with this is that humans can perceive things differently - either as several parts, or as these parts creating one whole. This can be seen in the laws of proximity, continuity and similarity. By looking at Figure 2.1 and 2.2 we can see that the dots and circles are individual elements, but we can also see them as groups. As for Figure 2.3, we can see two lines intersecting or we can see a “curved cross”.
It can also be seen in the law of figure/ground, albeit not as obviously. In Rubin’s vase (Figure 2.4), the figure can be the actual vase, and the ground is then the black background that it’s against. In another perspective, the figure can be two human faces in profile look-ing at each other, constructed from the contour of the vase, which now acts as a white (back)ground. The main point of the principle of figure/ground is that some objects may stick out more than others, which become the “ground”. The Rubin vase is a case of an ambiguous picture, where one can see different things depending on what one focuses on. However, this does not mean that every case of figure/ground (which exists naturally in our environment, not only as optical illusion images) is ambiguous. (Sternberg & Sternberg, 2011).
2.2. Methodology
2.2
Methodology
One cannot simply create something and then launch it for the general public to use without first testing and evaluating it to see how useful and usable the creation is. This study is only the first step of creating a grammar for a visual language, and even this first step needs to be tested and evaluated in iterations before the project can enter its next stage. The meth-ods used to evaluate the grammar are think aloud-tests complemented with semi-structured interviews. An analysis of the interviews was conducted to gather data that could be inter-pretable.
2.2.1
Think aloud test
According to Nielsen (2012) the think aloud method is very robust and has been used for several years. It is a very cost-efficient qualitative method, conducted in order to evaluate us-ability, where only five representative participants are needed. It is also very easy to conduct, all that needs to be done is find representative users, give them representative tasks to solve and lastly have the test leader sit down while listening and taking observational notes.
Nielsen (2012) mentions that other advantages of this method is that no special equipment is needed, and one can conduct think aloud tests at any testing point, from prototype to finished product or service. One disadvantage is that the participant is put in an unnatural situation - it is not overly common to verbalize one’s thoughts when doing tasks. However, Nielsen counters this by noting that participants are willing to do their best and that they often become so engrossed in the task that they forget that they are in a study.
Another disadvantage that Nielsen (2012) mentions is that sometimes the test leader needs to interrupt the participant by reminding them to think aloud, or inquiring on what they are thinking at the moment. This is not a problem per se, but depending on timing and wording this might affect the participant’s behaviour - leading to non-representative data. He urges the test leader to note which cases are like this and what impact the interruption may have had.
Nielsen (2012) also pushes that this is not a all method, but that it is useful in catch-ing what misconceptions the participant has of a system, and which elements needs to be changed - and most importantly, why they need to be changed.
2.2.2
Semi-structured interviews
Semi-structured interviews involve open questions and are more controlled than unstruc-tured interviews, since the interviewer will follow a set of scripted questions. They are also more informal than structured interviews, since the interviewer is allowed to clarify words and probe the interviewee about things they say. Hence, this is a suitable method for ex-ploratory and evaluational studies, since the questions and probes can be focused on trying to understand the interviewee’s attitude, values, thoughts and ideas (Louise Barriball & While, 1994).
Guest, Bunce, and Johnson (2006) conducted an experiment to see how many interviews are needed in order to get essential information. This is called saturation and is defined as “when the information given in the interviews no longer bring something new”, mean-ing when participants are saymean-ing the same thmean-ings as earlier participants and have no new ideas or feedback to contribute with. They explored this by conducting sixty in-depth, semi-structured interviews with women in West Africa. Their findings show that saturation occurs at about the twelfth interviewee, and that one should continue interviewing one or two par-ticipants to confirm that nothing new appears.
2.3. Related works
2.3
Related works
Humans have for a long time been interested in how to visualise information. This has re-sulted in attempts to create visual languages, some more general for communication and some more domain-specific. One of them is presented here. One of the foundations of the grammar in this study is the clinical terminology SNOMED CT, which is also explained at the beginning.
2.3.1
SNOMED CT
SNOMED CT is a comprehensive clinical health terminology product. The goal with this terminology is to standardize the the terms and phrases used in electronic health records. The terminology has been translated into several languages and is currently being used in more than 50 countries. It is also mapped to the medical classification list by the World Health Organization (WHO), in the tenth revision of International Statistical Classification of Diseases and Related Health Problems (shortened to ICD-10). There are visions for ICD-11 and SNOMED CT to share an ontology, thus creating ICD-11 with SNOMED CT in mind, rather than having to map everything a posteriori (What is SNOMED CT?, n.d.).
A terminology cannot “be used by itself”, hence why SNOMED CT must be implemented into an application or a software. It is useful when it comes to tasks and goals that include clinical documentation (How is NOMED CT used?, n.d.).
SNOMED CT is built on description logic and has a large ontology behind it. Ontologies are rules and guidelines that define and explain a system or “a world”. Few of these rules will be used in this study. These are SNOMED CT’s polyhierarchy and taxonomy, meaning the is-a (ISA) relis-ationships between concepts. This is is-a very common ontology, depicted by pis-arent- parent-child nodes that can be seen in a tree. An example of an ISA-relationship from SNOMED CT is that “pneumonia” is a child to both “lung consolidation” and “pneumonitis”, as well as being a parent to 36 children. The fact that a concept may have several parents and children also means that multiple inheritance is part of the ontology. Other rules includes categories such as body structure, disorder, procedure, regime/therapy and substance, as well as attributes (relationships between categories) such as finding site(1.4. SNOMED CT Features - Decision Support with SNOMED CT - SNOMED Confluence, n.d.).
Another part of the structure of SNOMED CT is that each clinical concept has a unique numeric identifier, connected to one preferred term and several synonyms, meaning that no matter which term a practitioner uses, it will connect back to the correct concept via the identifier. This is one part that represents the concept model of SNOMED CT.
Using these rules can help in guiding the creation of a grammar, especially since the goal is to in the end integrate the visual language with SNOMED CT.
2.3.2
Visualisation of Concepts in Medicine, VCM
One of the most prominent visual languages in the medical field today is VCM, created by Lamy, Duclos, Bar-Hen, Ouvrard, and Venot (2008). The focus of VCM lies in the pharma-ceutical field. The purpose of VCM was to create a graphical aid to drug monographs in order to minimise prescription errors, as well as shortening the time needed to find essential information. Instead of identifying a disease and then showing which medication should be prescribed, VCM focuses on showing which medication should be avoided, such as in case of allergies or bad chemistry with other drugs that the patient is taking.
The creation of the icons for VCM consisted of three steps, (a) creating graphically repre-sentable pictograms for the categories they needed (i.e. diseases, signs, physiological states and life habits, drugs and tests), (b) building a vocabulary and graphical primitives and (c) defining rules that govern how these primitives are allowed to be combined in order to create icons. Rather than having unique pictograms for the categories of diseases, drugs and tests
2.3. Related works
they decided to represent these by different types of attributes (e.g. anatomical location asso-ciated to the disease, drug or test, such as a brain for Alzheimer’s disease). This allowed the team to reuse icons across categories, using the same pictogram for different purposes. The actual drawing of the pictogram was made by following a schematic style with plain shapes (Lamy et al., 2008).
The grammar for VCM consists of the foundational components of the icons and how these components (graphical primitives) are allowed to be combined. These choices were made by the team with theories from cognitive science and from theories of human vision. The composition of icons also follows the principle of “is-a”, so if a concept for an icon has a parent-concept, that concept should somehow be visible in the child-concept. An example would be that heart failure is a type of cardiac disease, thus the icon for cardiac disease should be visible in the icon for heart failure. The basics of the icons include simple shapes (circle for normal patient state or square for pathological states), colours (brown for past/antecedent, red for current and orange for risks in the future), different kinds of shape modifiers to specify a drug or disease, attributes in the top-right corner (indicating drug or test) and lastly the pictograms which are usually anatomical or functional locations (Lamy et al., 2008).
When it comes to building sentences with these icons, Lamy and his team decided to use the “if-then” rule, such that “if disease then contraindication”. The “if-then” relationship is visualised through arrows and other geometrical shapes.
The first evaluation of VCM Lamy et al. (2008) was conducted to test its usability, how long time it would take to learn and to see whether it was understandable or not. Each of these components had its own test. The participants were 11 volunteer French general practitioners. The participants first had a meeting with the test leaders where they were introduced to VCM. They then received a CD-ROM with software on it, with the purpose of learning VCM. The participants were instructed to report how many hours they spent using the training software.
The evaluation test consisted of 100 icons or sentences that were presented to the prac-titioners. The participants were supposed to describe with natural language what the icons and sentences represented. The icons used in the test were chosen by the evaluation design-ers, with a range of simple to complex icons which covered many medical domains. The complexity of the icons were based on how many attributes an icon had (2 to 4 attributes) and how analogous the pictograms were (Lamy et al., 2008).
The usability test was conducted by letting the practitioners answer 20 yes/no-questions by deduction. Each question was associated with two documents - one in VCM and one in pure text. The participants got one or the other of the documents when answering questions, in a randomised order. The participants were timed during this test, to see whether they could find answers faster with VCM or not, and to see which type of document resulted in higher accuracy. Lastly, the participants evaluated themselves be answering if they replied faster and more accurately with VCM or without.
Quantitative statistical methods were used in this study. The specific tests used, with α=0.05, were:
• ANOVA, to investigate factors that could possibly affect the response times of the par-ticipants.
• Paired t-tests, to compare the mean response times.
• Fisher’s exact test, to compare number of errors between VCM and the pure text docu-ments.
• Linear regression, to observe the relationships between variables.
Overall the participants could correctly identify the icons and sentences with a mean of 87% and 75% respectively. Individually, however, the percentages of accuracy were between
2.3. Related works
65% to 99% for individual icons and between 24% to 100% for the sentences. Rgarding the yes/no-questions, 13 errors were made with VCM and 27 with the pure text, which resulted in a significant difference. The reading time was in general faster with the VCM icons, and more time was needed if more icons were present Lamy et al. (2008).
All in all, it was deemed a success and Jean-Baptiste Lamy and his co-founders have since then continued to work on VCM, going as far as to integrate it with the French quality-controlled health database CISMeF (Catalogage et l’Indexation des Sites Médicaux de langue Française) (Griffon et al., 2014).
Chapter 3
Method
The grammar was evaluated with think aloud tests and semi-structured interviews. Eleven out of thirteen of the tests were conducted in study rooms at the campus for faculty of medicine and health sciences at Linköping university. The other two tests were conducted over Skype, since the participants lived in other cities and could not do the tests in person. The sessions that were conducted in person went without problems and lasted between 20 to 50 minutes, depending on how well they understood the grammar and how much feedback they felt they could give during the interviews.
This chapter describes the procedure of creating the grammar and the icons, as well as how the grammar was evaluated. Demographic data about the participants is also presented.
3.1
Designing the grammar and icons
The design procedure of the grammar began by looking at the ontology of SNOMED CT and understanding the taxonomy, relations and other rules that are the core of the terminology. The CORE Subset helped in limiting which terms to observe and explore, since there are more than 350,000 concepts and terms in the terminology. The CORE Subset is a list of the most used terms in SNOMED CT, with the statistics gathered from eight institutions. After acquiring a better understanding of SNOMED CT and having read about gestalt laws and VCM it was time to creatively and boundlessly sketch ideas on what a grammar can look like, i.e. how to place icons in a logical order, or even without any order at all, as well as exploring which shapes and colours that can be used. After having sketched several ideas it was time to choose one of the concepts to develop. At this point it is needed to somehow represent the grammar - icons were needed.
It is difficult to evaluate a grammar without somehow visualising it. For this study, six terms from the CORE Subset were used: hypertension (disorder), hyperlipidemia (disorder), de-pressive disorder (disorder), adult health examination (procedure), diabetes mellitus (disorder) and coronary arteriosclerosis. Other terms that were related to these six were designed as well, e.g. nutrition therapy and take blood sample in order to be able to visualise the grammar. All icons can be found in Appendix A.
The grammar, icons and the test were created with the illustration software KRITA version 3.1.2. The hardware used was a Wacom Volito drawing tablet and an ASUS Notebook X552C.
3.2
Participants
The people who were eligible as participants were individuals who work in the medical field as doctors, nurses or general practitioners, as well as university students aspiring to become doctors or nurses themselves. No limit was put on age or work experience/number of semesters studied.
3.3. Test structure
Thirteen individuals participated in the study and everyone completed the test and inter-view. See table 3.1 for demographic information. Eleven of them did the test in person, while the other two did it via video call (Skype).
Table 3.1: Participant demographics.
No. of participants Age Occupation Semester (Male, Female)
Range: 21-57 Doctor 3 (3, 0)
Median: 26 Medical student: doctor Range: 3-11 7 (2, 5) Mean: 32.8 Medical student: nurse Range: 1-6 3 (0, 3)
Total:13
3.3
Test structure
The test was put together with the following documents: (a) pictures of the grammar and icons (see Figure 3.1, Figure 3.2) (b) the instructions (see Appendix B) (c) a legend describing the bases of the icons (see Appendix C), (d) an informed consent document and (e) interview questions, which were as follows (translated from Swedish):
1. What do you think of the placement of the icons?
2. What do you think about the shapes and colours of the icons bases? 3. What was most clear about the structure?
4. What was least clear?
5. What would you like to have added in order to easier understand the information? 6. How detailed do you believe that visual information can be in patient journals?
Two pilot tests were conducted to see whether the procedure of the test was logical and easy to follow, and if enough information was given in the instructions and informed consent. A few changes were made, such as:
• Replacing titles with “Patient 1” and “Patient 2”. The titles signified each column (from left to right, Reason for visit, Procedure, Diagnosis and Treatment.
• Changing the pictogram for "blood sample".
• Expanding the distance between the two icons that represented "adult health examina-tion".
• Removing the superfluous interview question of "Was something more difficult to un-derstand? If yes, what?"
3.3. Test structure
Figure 3.1: The first page of the test with the grammar and icons.
3.4. Test procedure
3.4
Test procedure
Invitations were sent out via email after the pilot tests were done. The contacted people were to book a time via the invitation.
At the beginning of each session the test leader explained the purpose and motivation of the study. The participant received an informed consent to sign. Then the participant received a paper with the instructions, ordered into three steps. The steps were to be read one at a time. Step one explained the legend. After this step was read by the participant, they received the legend document and were allowed to look at it and ask questions. This document was to be placed face-down on the table and the participant could look at it at any given time.
Step two explained the actual test. After this step was read the participant received the two papers with the grammar and icons, and they could begin with the think aloud test.
Step three explained the interview and the questions were asked as soon as the participant was done with step two. At this point they were also informed that the icons would be ex-plained after the interview questions. After the sessions the participants were contacted again to be asked if they were open to participate in future developmental stages of the project.
Chapter 4
Results
In this chapter the grammar is presented and explained, as well as excerpts from the inter-views.
4.1
The grammar
Figure 4.1: The basic shapes and colours for the grammar.
Four categories were created in order to structure and group concepts from SNOMED CT. These are anatomical location, procedure, diagnosis and treatment. Basic shapes and colours were created to represent these. The shapes and colours are a blue square, a turquoise rhombus, a yellow triangle and a green circle. These are presented in Figure 4.1. The groupings and mappings from SNOMED CT categories to the grammar categories can be seen in Table 4.1. The idea of having unique shapes and colours for categories is inspired from the VCM icons.
Table 4.1: Category mapping.
SNOMED CT The grammar Shape and colour
Body structure, finding site Anatomical location Blue square
Disorder Diagnosis Yellow triangle
Procedure Procedure Turquoise rhombus
Regime/therapy, substance Treatment Green circle
Figure 4.2:
Up-and
down-arrows. Concepts from SNOMED CT can be represented either by one icon or
by two. When two icons are compounded to represent a concept, they are connected by a black line (see Appendix A for all icons). An example is given in Figure 4.3. This black line also follows the principle of continuity. A green arrow is also part of the compound icons in Figure 4.3 - this is an auxiliary element. As of now in the grammar, there only exists three types of auxiliary elements. The first two are the green up-arrow and the red-down arrow, representing “high/increase” respectively “low/decrease”
(see Figure 4.2). The third one is textual acronyms for words, such as “CAD” = coronary artery disease, “LDL” = low density lipids and “INS” = insulin.
4.1. The grammar
Figure 4.3: Left: Compounded icons to represent high blood pressure. Right: One icon that represents hypercholesterolemia.
An example placement of icons can be seen in Figure 4.4. Four implicit columns are cre-ated by having icons grouped together by proximity. Another grouping is present through proximity together with similarity, where icons of the same shape and colour are put close together. These columns indicate an implicit time axis, since they can from left to right be seen as Reason for visit, Procedure/how to find out the problem, Problem/diagnosis and treatment. Similarity can also be seen by each “patient visit”, meaning that the icons come in the same order for each entry.
The linear placement is partly due to the “if-then” placement of VCM icons, and partly due to that Westerners read from left to right, and lastly because the icons follow an implicit time axis.
Figure 4.4: Example of the placement of icons. Grouping by proximity and similarity by icon shape and colour.
The pictograms for the icons were based on conventional pictures that have been widely used throughout history (e.g. cutlery for “food”), one exception being the international sym-bol for diabetes (a blue circle). Base pictograms were used to show similarity and the parent-child (ISA) rule from the ontology of SNOMED CT. This makes it possible to generalise the icons in order to reduce the amount of unique icons. An example is shown in Figure 4.5, de-picting that coronary artery disease and hypercholesterolemia are types of vascular diseases, indicated by the base pictogram of “blood vessel”.
Figure 4.5: Example of use of the same base pictogram.
The principle of figure/ground is visualised partly by the the pictogram against the base of the icons, and partly of the icon itself against the background. The icons were designed with a white background in mind. Using black lines rather than lineless pictograms is a conscious decision to mitigate the problem of the foreground colours blending into the back-ground colour.
4.2. The interviews
4.2
The interviews
All thirteen participants answered all of the interview questions. All of them came with thoughts and ideas on how to improve comprehension of the grammar and icons. Here, some recurrent and unique feedback will be brought up, which will later be discussed.
All of the participants were positive to the general idea of having complementary icons to patient journals. Several of them commented that it would decrease the amount of time and effort to find essential information if icons existed to give an overview of a patient. It is also important to note that some participants voiced their concerns on how to actually use the icons - if it meant writing personal notes for each icon as annotations then that simply creates more work for doctors, nurses and medical assistants. If this is the case, then these participants would regard it as an unhelpful system.
Four of the participants commented that it felt like following a story when doing the test, even if they realised this after having begun deciphering the individual icons. One mentioned that it felt easier to comprehend the rest of the icons after having gotten this perspective. Another commented that they would have understood that there was a flow earlier if there was an explicit time axis. This was also important to this participant since they thought about non-Westerners, who do not read from left to right.
One participant mentioned that it might be worth having two “modes” for the grammar - one that represents a certain patient visit and one that represents the current status of the patient. They commented that this would make it more clear and structured in what earlier diagnoses and treatments the patient has had, and what diagnosis and treatment they are cur-rently having. This participant also explained and illustrated their idea, more details about this under 5.1 Results.
Another suggestion from several participants, with regards to the interview questions “What would you like to have added in order to easier understand the information?” and “How detailed do you believe that visual information can be in patient journals?”, is to have pop-up messages appear when the user hovers over or clicks on an icon. This pop-up mes-sage would then have more specific information in text and parameter values, such as the exact diagnosis or medication (one of the doctors commented that this can be connected to the preferred term in SNOMED CT via its concept ID), amount of the medication prescribed, date/time span, optional personal notes from practitioners and/or hyperlink to the relevant free text part of the journal. An alternative to the last part is to “reverse” it by having relevant text from the journal appear in the pop-up message. Participants who were presented with this idea during the interview agreed that it would be a good idea.
None of the participants understood the meaning behind the symbol for diabetes, and when asked about what the least clear thing was about the structure almost everyone in-cluded this symbol. One participant even suggested that it might be better to make it look more literal, such as having a pictogram of a drop of blood with a sugar cube in it.
Many also included the auxuliary acronyms as being difficult to understand. The acronyms for this study were taken from English, and only part of the participants knew or remembered what “CAD” was an acronym for (in Swedish, LDL and INS would have the same acronym). One participant also commented that the same acronyms may be used between medical fields - meaning that they have the same acronym, but different meanings.
One other suggestion is that of changing the colours of the auxiliary up- and down-arrows. The design as of now is a green up-arrow and a red down-arrow. All of the par-ticipants realised that the green up-arrow indicated high blood pressure in the context, but noted that it is unclear whether it means if it is an improvement or not. After all, “increase” is not equal to “improvement” and the same goes for “decrease” and “worsening”. Three participants also mentioned that they will most likely be problematic for people who are red-green colourblind.
Some participants commented that they did not understand why the icon for “depressive disorder” appeared twice in the same row. They understood after the interviewer explained
4.2. The interviews
the reasoning, and several participants suggested creating another base icon (i.e. shape and colour) in order to differentiate between “problem” and “diagnosis”.
Two of the doctors mentioned a project that was somewhat similar in aim as this grammar. This was the attentional information symbol (Hoffmann & Lövström, 2011; Lövström, 2008). They both noted that it might be worth having contemporary work in mind when developing this grammar.
Chapter 5
Discussion
This is, as earlier mentioned, only the first step in creating a grammar for visualising and summarising information in electronical health records. Thus, there lies much potential in developing the components of it and many improvements can be made. All of the partici-pants in the study contributed with new ideas and concepts. The major ones, and the ones that several participants agreed on, will be discussed here.
This project of creating and implementing a visual language will have to go through sev-eral iterations to be as useful and as usable as possible. This is why the method of this study will be discussed - how the different parts were conducted and the advantages and disad-vantages of them. This discussion comes after the discussion of the results.
5.1
Results
Most of the participants were able to follow the logic behind the grammar, some even regard-ing it as “followregard-ing a story”. However, not all of them understood that all icons in each row belonged to one visit or even one patient. One interesting observation is that participants who at first did not see the icons in one context realised this after noticing the dates at the be-ginning of each row. At this point they seemed to have become enlightened and backtracked through the icons that they had already analysed, seeing them in a new light. When this oc-curred, some icons were re-evaluated and in other cases the participant simply realised that it was one visit rather than random icons (in one way they had “taken a step back” and saw it at a higher level). The same phenomena happened to some participants that would analyse the icons in order, but could not come to a conclusion until they looked at the ones that came afterwards. Here, they would use the additional information that they understood in order to deduce the past icon. This happened most often with the icons representing diabetes and depression - only after the participant saw the treatment did they understand what the di-agnosis was. This means that context somehow needs to be established first, before the user starts interpreting the icons.
This is a problem since the idea of this visual language is that it is supposed to decrease misunderstandings and speed up the process of becoming updated on a patient’s status. One of the interview questions were “What would you need to be able to easier understand the information?”. Many answered that they probably would have understood the context if there somewhere was an explicit time axis, that would indicate that the icons are part of a flow. One of the participants knew a language where text is read from right to left - they commented that for people who were not multilingual in that case could possibly become confused. Nevertheless, they also believed that it would not be a problem if a time axis was incorporated - as long as the reader has a starting point in interpreting the icons. Even linearity and order is not fully necessary in order to understand the icons, however, if they are not obvious then it takes a longer time for the reader to recognise their meaning. Other ways of experimenting with icon order is to automatically reverse the order for languages that are read from right to left, or put them on a vertical axis since all languages are read
5.1. Results
from top to bottom (Kolers, 1969). It is also worth noting that some pictograms may possibly have to look different depending on language - is it natural and understandable to have cutlery to represent diet/nutrition/food when used in a country which predominantly uses chopsticks? Do all countries have access to the same kinds of medical equipment? Language is a major part of this study, but we cannot remove culture. Language affects culture and how we see things, but culture also affects language and which knowledge we possess. This in turn affects how we may interpret pictures (Danesi, 2007; Kolers, 1969).
During the interviews many suggestions and solutions were brought forward, both for the grammar and the pictograms. Thus, it would be ideal to have workshops together with graphical designers and the end users to further develop the grammar and the pictograms in the icons. Notable suggestions include having two “modes” of the grammar - one that represents a patient visit (the grammar as it is now) and one for summarising the patient’s current state. The first mode would be static and archived, while the other one would be kept updated in new entries of the patient’s journal. Differences in the grammar would be that in mode no. 2 the icons for all the procedures conducted would be grouped in a vertical line, then one or several (set or assumed) diagnoses would come, and after that the current treatment. This would be useful since there are many ways to treat the same problem, and it depends on the individual for what works best - meaning that changes in treatment (and assumed diagnosis!) may change from time to time. Mode no. 1 would be, as mentioned, archived. This would make it possible to backtrack and see which procedures have been done, which diagnoses have been set in the past and which treatments have been tested before.
Another suggestion brought up was that of implementing pop-up messages. This would decrease the amount of unique icons that would be needed to memorise, hence making way for a more generalised visual language. The visual language is, after all, supposed to be complementary to the free text and its purpose is to give the reader an overview. If the icons are too many and too detailed, it would instead increase cognitive workload rather than decrease it. This can be compared to the alphabet and the binary system - learn the letters or meaning of the numbers and you have the basis to learn many different languages and compound many different numbers by using few key elements. However, if you were to learn Chinese or the decimal system, you suddenly need to learn the individual meaning of each symbol, and instead of learning two elements (0 and 1 in binary) you suddenly need to learn 10 elements in order to be able to compound numbers. The consequence of the latter is that the result is much more compact (compare 720 to 101101000)(Kolers, 1969). Here, it becomes a trade-off - is it worth learning hundreds, maybe thousands of more specific icons, or is it more effective to learn less, more general icons?
When it comes to interpreting and memorising the pictograms in the icons it might be best to keep them as literal as possible - partly to speed up recognisation and partly because symbols may not be used internationally (Danesi, 2007). An example of when this became a problem is the icon of diabetes mellitus - the yellow triangle representing a problem or disease, with the pictogram of a simple blue cirlce. The blue circle is, in some countries, regarded as the “international symbol for diabetes”. However, not one of the participants knew what the blue circle was supposed to mean. A few participants had thoughts along the same lines, that it might mean “general problem” or that it was the number 0, which might mean that there is a failure of the pancreas and/or kidneys (the anatomical location that the diabetes-icon was connected to). One participant suggested to make the pictogram more literal, such as having a drawing of a drop of blood with a sugar cube in it.
The problems of not understanding the auxiliary text for some icons is in accordance to Chute (2005), where he writes that the same acronyms with different abbreviations span across domains. The earlier mentioned idea of pop-up messages would also eliminate this problem, since there would be no need for auxiliary text that in some cases may confuse rather than clarify. It is important to note that some participants liked the text and felt that
5.2. Method
they could gather more information than with just the pictogram, but agreed that they would be superfluous if the pop-up messages were implemented.
One argument against green and red arrows by three participants was that it would not do well with people who had red-green colourblindness. A solution for this was to change the red and green colour depending on context (say, if high blood pressure in a case would be regarded as an improvement, the up-arrow would be green. If it instead would be a bad sign, it would still be an up-arrow, but red). On the other hand, this contradicts the argument of colourblindness. Other solutions were either to give both arrows a neutral colour (i.e. black), or have one arrow be black and the other gray or white. Personally I would advocate the use of the same colour for both arrows, and let the context or pop-up messages indicate whether it is an improvement or a worsening. This is motivated by the fact that the same problem of “increase” versus “improvement” could occur if the arrows were to continue being differently saturated - it is better to strive for as non-ambiguous icons as possible (Kolers, 1969).
With regards to ambiguity, several participants were confused as to why the depressive disorder-icon appeared twice in the same entry. Regarding this, and the fact there is a differ-ence between “problem” and “disorder”, led some participants to suggest creating another icon shape with corresponding colour to represent either of those, rather than putting both under “diagnosis”. The main idea is to separate problem and disorder. The difference would be that a problem would usually be from the patient’s perspective - they encounter a prob-lem and visit a doctor to find out more. During and after the visit it becomes a diagnosis - it turned out to be a disorder, or the doctor is unsure and keeps it defined as a problem until further tests have been done.
Lastly, it is important to keep in mind of similar contemporary works. Two participants mentioned a project where an attentional information symbol has been developed to show-case warnings about a patient (Hoffmann & Lövström, 2011; Lövström, 2008). The symbol is based on the asterisk shape, often associated with the medical field together with the rod of Asclepius. The middle column of the shape represents an exclamation mark, which marks the severity level of how sensitive a patient is to certain medication (with three levels), the top left “arm” represents unstructured information, the bottom left is yellow and represents contagiousness, the bottom right is blue and represents deviation in treatment and the top right represents diagnoses and treatments. This project also suggests using the symbol as an overview, and getting more detailed information through text by clicking on the symbol or the relevant part of it. The visualisation is however slightly different - rather than having pop-up messages with text, clicking on the symbol would change page for the user in the patient journal. In his report, Lövström (2008) writes about an interview with Professor Uvo Hölscher, who mentions that it may be impossible to create a truly international symbol that will be understood by everyone. I agree that it seems implausible, but as mentioned earlier in this chapter (when discussing linguistic relativity) I believe that we will need to go around this by having some pictograms be local.
5.2
Method
The first Skype-session lasted about an hour and could not be recorded due to technical prob-lems. The second Skype-session was recorded as planned, however the connection was dis-rupted two times, once during the test and once during the interview. This sessions lasted more than an hour due to these disruptions. Both participants also had slight difficulties in hearing the test leader, possibly due to the test leader’s microphone. These deviations and disruptions were not regarded as major problems, since the test leader meticulously took notes and made sure to note recurring themes and keywords that participants before them had mentioned.
One constant issue with recording users is that this may affect their behaviour and perfor-mance. In most cases this is not a problem, since the participant quickly becomes engrossed
5.2. Method
in the task and forgets that they are being observed and recorded. All the participants in this study seemed to have reacted this way and no one seemed to be uncomfortable.
The test was conducted traditionally with pen and paper. Some of the participants com-mented on this, noting that the icons were fine in the size that they were on paper, but were wondering if it would have been the same on a digital PC-screen. The icons are supposed to be used mainly digitally, so after revising the grammar and icons from this study it would be necessary to also test it digitally. In interaction and user experience design it is nevertheless very common to begin with paper prototypes, especially when working in iterations. This is due to paper prototyping being very cost-effective. The test being on paper also allows the participants to draw and write comments where it is appropriate.
These types of projects, i.e. design projects in general, are best done in teams. I did this alone due to it being my bachelor’s thesis. However, creating things from scratch, from the concept phase, need the involvement of several perspectives and opinions. Future develop-ment on this project should thus be done with several people of both the same and different competencies, in order to create something useful and usable. It is also important to include the end user during the development in order to create a good service from the beginning, rather than having to spend time and resources to touch it up after its completion.
Chapter 6
Conclusion
This study aimed to create and evaluate a grammar for a visual language to be used in the medical field. The goal of this study was to answer the questions:
1. How can a grammar for a visual language look like when it is partly based on the ontology of SNOMED CT?
2. What level of detail of a visual language is appropriate for people in the medical field when reading patient journals?
3. Which information do people in the medical field need in order to understand a medical visual language?
The first question can be answered by looking at the basis of the icons that were designed, and in which order they are put. Each category has a specific shape and colour assigned, with pictograms in each icon. Auxiliary items such as an up- and down-arrow exist in order to depict “increase/decrease” or “improvement/worsening”. These elements together create a full icon, which by itself or compounded with other icons through a black line represents clinical concepts from SNOMED CT. Other aspects from the ontology is that the same base pictograms are used to represent taxonomy, and connecting individual icons with a black line to represent new terms and, in some cases, the attribute of finding site. These elements and how they are put together stemmed from gestalt laws and part of the ontology behind SNOMED CT. Inspiration for some aspects were taken from VCM.
The second question can be answered by summarising the answers from the interviews. There were some mixed opinions on the level of detail, but most participants agreed that it would be more useful to have the icons be more general in nature and have more specific information as text. One way to access this information would be to insert the relevant text snippets and other parameter values in pop-up messages that would appear when the user hovered over or clicked on an icon. The reason to keep it more general is to decrease the amount of unique icons (or rather, pictograms) that the user would have to learn, and also because the purpose of this visual language is to act as a complement. Since it is a com-plement, it does not need to be fully specified. The article by Kolers (1969) reinforces this argument.
The third question can also be answered by summarising answers from the interview questions. One idea is to create two “modes” in order to more clearly be able to represent a patient’s current status and what has been done to the patient earlier. Another idea is to have an explicit time axis, both to indicate that the icons are part of a flow and are related to each other, but also to create a starting of where the reader is supposed to start. This would mitigate the problem of whether people read from left-to-right or vice versa. Another suggestion was to discard auxiliary text and instead have more specific information in the journal, which quickly can be accessed via pop-up messages.
As mentioned earlier, this is merely a first step in a big project. There is potential for many improvements and for future development it is important to work in iterations and in teams - include the end users in the process and create a usable system from the beginning.
Managing patient journals and other electronical health records is not an easy task and errors may result in the loss of life - let us work together to make the managing of medical documents more efficient for everyone involved, in order to save lives more efficiently.
References
1.4. SNOMED CT Features - Decision Support with SNOMED CT - SNOMED Conflu-ence. (n.d.). Retrieved from https://confluConflu-ence.ihtsdotools.org/display/ DOCCDS/1.4.+SNOMED+CT+Features
Chute, C. G. (2005). Medical Concept Representation. In C. Hsinchun, S. F. Sherri-lynne, C. Friedman, & W. Hersh (Eds.), Medical informatics. knowledge management and data mining in biomedicine. (1st ed., Vol. 8, pp. 163–182). Springer US. doi: 10.1007/ 0-387-25739-x{\_}6
Danesi, M. (2007). The Quest for Meaning: A Guide to Semiotic Theory and Practice (M. Danesi, U. Eco, P. Perron, R. Posner, & P. Schulz, Eds.). Toronto: University of Toronto Press. Griffon, N., Kerdelhué, G., Hamek, S., Hassler, S., Boog, C., Lamy, J.-B., . . . Darmoni, S. J.
(2014, 10). Design and usability study of an iconic user interface to ease information retrieval of medical guidelines. Journal of the American Medical Informatics Association : JAMIA, 21(e2), 270–7. doi: 10.1136/amiajnl-2012-001548
Guest, G., Bunce, A., & Johnson, L. (2006). How Many Interviews Are Enough? An Experiment with Data Saturation and Variability. Field Methods, 18(1), 59–82. doi: 10.1177/1525822X05279903
Heider, G. M. (1977). More about Hull and Koffka. American Scientist, 32(5), 383. doi: doi:10.1037/0003-066X.32.5.383.a
Hoffmann, M., & Lövström, R. (2011, 11). Livsviktigt symbol ställer krav. Läkartidningen No. 48 Vol. 108, 2472–2473. Retrieved from http://ww2.lakartidningen.se/store/ articlepdf/1/17413/LKT1148s2472_2473.pdf
How is NOMED CT used? (n.d.). Retrieved from http://www.snomed.org/snomed-ct/ what-is-snomed-ct/how-is-snomed-ct-used
Kolers, P. A. (1969). Some Formal Characteristics of Pictograms. American Scientist, 53(3), 348–363.
Lamy, J.-B., Duclos, C., Bar-Hen, A., Ouvrard, P., & Venot, A. (2008, 4). An iconic language for the graphical representation of medical concepts. BMC medical informatics and decision making, 8, 16. doi: 10.1186/1472-6947-8-16
Louise Barriball, K., & While, A. (1994). Collecting data using a semi-structured interview: a discussion paper. Journal of Advanced Nursing, 19(2), 328–335. doi: 10.1111/j.1365-2648 .1994.tb01088.x
Lövström, R. (2008). Varningsinformation Etapp 2: Slutrapport (Tech. Rep.). Sjukvårdsråd-givningen. Retrieved from http://www.nepi.net/res/dokument/Slutrapport -varningsinformation-etapp2.pdf
Nelson, D. L., Reed, V. S., & Walling, J. R. (1976). Pictorial Superiority Effect. Journal of Experimental Psychology: Human Learning and Memory, 2(5), 523–528. doi: 10.1037/0278 -7393.2.5.523
Nielsen, J. (2012). Thinking Aloud: The #1 Usability Tool. Retrieved from https:// www.nngroup.com/articles/thinking-aloud-the-1-usability-tool/ Sternberg, R. J., & Sternberg, K. (2011). Cognition (6th ed.). Belmont: Wadsworth.
Wertheimer, M. (1938). Laws of Organizations in Perceptual Forms. In W. D. Ellis (Ed.), A source book og gestalt psychology (pp. 71–88). Pennsylvania: Gestalt Legacy Press.
References
Retrieved from http://psychclassics.yorku.ca/Wertheimer/Forms/forms .htm
What is SNOMED CT? (n.d.). Retrieved from http://www.snomed.org/snomed-ct/ what-is-snomed-ct
Wiedenbeck, S. (1999). The use of icons and labels in an end user application program: An empirical study of learning and retention. Behaviour & Information Technology, 18(2), 68–82. doi: 10.1080/014492999119129
Appendix A
Appendix B
Test Instructions
Steg 1
I det här testet kommer du att få se ikoner som, enskilt eller tillsammans med andra ikoner, representerar en medicinsk term. Innan testet påbörjas kommer du att få ett papper som förklarar vad de olika formerna betyder, samt ett par exempel för att bekanta dig med ut-seendet. Det här pappret kommer du sedan kunna be om att få se igen under själva testet, om du känner att du skulle behöva den hjälpen. Nu kan du ställa eventuella frågor om testet till forskningsledaren.
Steg 2
Nu kommer du att få två ihophäftade papper. Detta är själva testet. Ikonerna kommer att vara utställda på ett visst sätt med streck under. På strecken ska du skriva vilken term du tror att ikonen/ikonerna representerar. Det finns ett streck per term. Om du inte kom-mer på en viss term så går det bra att beskriva ordet. Du kan skriva på svenska eller engelska. När du gör testet ska du tänka högt - detta betyder att du säger allt som du tänker på i stunden. Var inte orolig om det du tänker på är relevant eller inte, allt du tänker på är viktigt. Om du glömmer bort att prata så kommer forskningsledaren att påminna dig.
Det är inte du eller din kunskap som testas. Det är heller inte själva ikonerna som tes-tas, utan det är placeringen av dem.
Steg 3
Nu kommer forskningsledaren att ställa ett par frågor till dig angående testets innehåll. Du behöver inte svara på frågorna om du inte känner för det.
