Guideline Overview Tool (GOT)

Asgaard-TR-2001-4, 2001 Guideline Overview Tool (GOT) Content 1 Field of Application and Purpose 1.1 User 1.2 Input / Output Devices 1.3 Internationalisation 2 Investigation of Software and Concepts Wolfgang Aigner Vienna University of Technology Institute of Software Technology and Interactive Systems (IFS) email: aigner@asgaard.tuwien.ac.at phone: +43-1-58801-18833 fax: +43-1-58801-18899 www: http://www.asgaard.tuwien.ac.at/~aigner/ 2.1 Overview, method 2.2 Search for Commercial Medical Software 2.3 Evaluation of Visualizations in Tools for Project Management 2.4 Search for Possibilities to Represent Temporal Processes Graphically 2.5 Summary 3 Visualization of Data Items and their Relations 3.1 Visualization of Plans 3.2 Visualization of Values 3.3 Time Cursor 3.4 Time Scale 3.5 Panels and their Order 3.6 Putting it all Together 3.7 Example 4 SQUEAK Prototype 4.1 Introduction 4.2 Prototype Features 4.3 Testing the prototype 4.4 Structure of UI parts

Field of Application, Main Purpose 8 June 2001 Guideline Overview Tool: Field of Application, Main Purpose The aim of the Clinical Guideline Overview Tool is to support physicians in executing therapies with clinical guidelines. This tool is designed for execution support only and not for the creation of clinical guidelines like the already existing tool AsbruView. The main purpose of this software is to provide a compact overview that is easy to read and understand. This overview has two functions: At the one hand side it should allow a comparison of the actual state of a patient and the assigned clinical guideline. On the other hand side it should be possible to compare several patients at one view. This view includes at most five patients synchronously. 1. Users The users of this tool will be physicians. Nursing personnel is explicitly excluded in the design because in this case substantially different points of interest would have to be considered. Further we assume that the users do have a certain background in dealing with computer software. This means that he/she has practical experience in the use of a window-oriented operating system like MS Windows or Mac OS. 2. Input / Output Devices As output device a monitor with a size of at least 17 and a resolution of 1024 x 768 at minimum should be used. Inputs will be made via mouse and keyboard. These preconditions are crucial for the design of the size of user interface elements. 3. Internationalization The graphic design of this tool is developed for the western cultural area. This restriction is important for the use of colors, icons and metaphors at the user interface. The meanings of these parts differ widely among various cultures. Therefore we focus on the western culture. Author: Wolfgang Aigner, 9755342 Page: 1 / 1

Investigation of software and concepts 8 June 2001 Guideline Overview Tool: Investigation of software and concepts 1. Overview, method The search for available software and concepts in the area of the comparative visualization of medical therapy plans took basically place in the following areas: Search for commercial medical software Evaluation of visualizations in tools for project management Search for possibilities to represent temporal processes graphically 2. Search for commercial medical software In this area the search mainly took place on the basis of a listing of medical software manufacturers at yahoo: Yahoo! Business and Economy > Companies > Health > Software > Health Care Management (http://english.hk.yahoo.com/business_and_economy/companies/health/software/health_ca re_management/ ) The products offered by these companies mainly deal with the administration of individual patients and their data but they don't offer comparisons between different patients. In addition only very few of this software products offer any diagrams or graphical representations. Further emphasis is given to software products for the administration of medical practices (financial, medicines...) and to the supply of services via Internet. There were neither products dealing with skeletal medical plans within the examined companies nor their software visualized comparisons of any kind. 3. Evaluation of visualizations in tools for project management The main types of representation used in tools for project management are GANTT or PERT charts. But these types of diagrams are not suitable for the comparative representation of medical skeletal plans because of the following reasons: they require too much vertical space for a comparative representation repeating and any order plans are not representable the degree of detail of the representation can not be altered For the Tool AsbruView of the Asgaard project the use of PERT or Gantt diagrams has already been examined: [1] Author: Wolfgang Aigner, 9755342 Page: 1 / 3

Investigation of software and concepts 8 June 2001 4. Search for possibilities to represent temporal processes graphically Afterwards I looked up visualizations within the area of medicine and graphical comparisons of plans. The site Health InfoDesign (http://www.healthinfodesign.com) contains rather general tips for the design of user interfaces within the medical area. It also offers some interesting links and tips for books. In the e-zine of the human computer interaction designer Bruce Tognazzini AskTog (http://www.asktog.com) you can find a huge amount of useful design tips as well as criticisms of more or less bad user interfaces. The author worked at Apple and Sun and is now chief designer at Healtheon/WebMD, a company for medical software and web services. However probably the most interesting site I found was that of the Human Computer Interface Lab (HCIL) (http://www.cs.umd.edu/hcil) of the University of Maryland. That is one of the leading research labs in the area of human computer interaction with its head Ben Shneiderman. They do several projects concerning new concepts in the area of HCI and introduce them at this site. For our project the most relevant idea is that of the use of LifeLines for the visualization of patient data ([2], [3]). This project deals with a representation of patient documents (personal data, drugs, diseases, etc..). There are neither therapy plans nor possibilities of comparisons offered. Nevertheless this type of representation seems to be also very promisingly to our problem. Starting with the papers of the LifeLines project of the HCIL I looked for further documents dealing with this topic at the ResearchIndex of the NECI Scientific Literature Digital Library CiteSeer (http://citeseer.nj.nec.com) and found some interesting texts:[5], [6]. 5. Summary I spent very much time to examine the commercial products of manufacturers of medical software and didn't find one that represents different patients at one view, offers comparisons between patients or deals with skeletal plans. The types of representations used in the area of project management are not applicable for our purposes due to the given reasons. The only promising idea I discovered is the representation by LifeLines. One can conclude from this result that the project of the comparative visualization of medical therapy plans is very necessary, since nothing comparable exists within this area. Author: Wolfgang Aigner, 9755342 Page: 2 / 3

Investigation of software and concepts 8 June 2001 6. Bibliography [1] Robert Kosara and Silvia Miksch. A Visualization of Medical Therapy Plans compared to Gantt and PERT Charts. Vienna University of Technology. Institute of Software Technology (IFS). Favoritenstraße 9-11/E188 Wien A 1040, http://www/ifs.tuwien.ac.at/asgaard/, < rkosara@ifs.tuwien.ac.at>, <silvia@ifs.tuwien.ac.at>. In Scott Goodwin and Andre Trudel (eds), Proceedings of the Seventh International Workshop on Temporal Representation and Reasoning (TIME- 00), IEEE Computer Society, pp. 153-181, 2000. [2] Catherine Plaisant and Anne Rose. Exploring LifeLines to Visualize Patient Records. University of Maryland. Human-Computer Interaction Laboratory. A.V. Williams Bldg. College Park MD 20782, http://www/cs.umd.edu/projects/hcil, <plaisant@cs.umd.edu>, <rose@cs.umd.edu>, Technical Reports CS-TR-3620 (1997). [3] PhD. Catherine Plaisant, PhD. Richard Mushlin, MD. Aaron Snyder, Jia Li, Dan Heller, and PhD. Ben Shneiderman. LifeLines: Using Visualization to Enhance Navigation and Analysis of Patient Records. University of Maryland. Human-Computer Interaction Laboratory. A.V. Williams Bldg. College Park MD 20782,. IBM T.J. Watson Research Center. Kaiser Permanente Colorado. In Proceedings of the 1998 American Medical Informatic Association Annual Fall Symposium, pages 76-80, 1998. [4] Yuval Shavar and Cleve Cheng. Knowledge-Based Visualization and Navigation of Time- Oriented Clinical Data and their Abstractions. In IDAMAP 1998, Brighton, UK,. 1998. [5] Mark Derthick and Steven F. Roth. Data Exploration across Temporal Contexts. Carnegie Mellon University. Robotics Institute. http://www/cs.umd.edu/projects/hcil, <mad@cs.cmu.ed>, < roth@cs.cmu.edu>. In Proceedings of Intelligent User Interfaces (IUI '00), New Orleans, LA. pp. 60-67. [6] Stephan Greene, Gary Marchionini, Catherine Plaisant, and Ben Shneiderman. Previews and Overviews in Digital Libraries: Designing Surrogates to Support Visual Information Seeking. University of Maryland. Human-Computer Interaction Laboratory. A.V. Williams Bldg. College Park MD 20782,. College of Computer Science. Institute for Systems Research, Technical Report CS-TR-3838, 1997. Author: Wolfgang Aigner, 9755342 Page: 3 / 3

Guideline Overview Tool: 1. Visualization of plans For the visualization of plans we use LifeLines because of their powerful possibilities and little vertical space consumption. For the reason of displaying parallel processes we expanded the model of LifeLines by using several vertically shifted line segments. Thus line segments one above the other aren t several LifeLines, they are belonging to the same LifeLine. Plan states In Asbru there are various states of a plan as described in the Asbru 7.2 Reference Manual [1] at page 72. For the comparative visualization the various plan-execution states are displayed. As another parallel view, also the plans in the state possible can be displayed. activated: A state transition to the plan state activated is shown by the sharp beginning of a bar for that plan. While a plan is activated it is shown as filled bar: activated suspended A transition of a plan to the state suspended is only possible via the state activated. Such a transition is displayed by a zigzag line. If a plan holds the state suspended it is displayed by a non-filled bar with dotted edges: Author: Wolfgang Aigner, 9755342 Page: 1 / 11

suspended aborted A transition of a plan to the state aborted is displayed by the sharp end of the bar and a stop sign within the bar (red octagon with white letters STOP ). A plan can be aborted via the states active and suspended. If there is too little space to draw a stop sign it may be feasible to mark the end of the bar with a thick red line: or: aborted completed The sharp end of the plan bar shows the transition of a plan to the state completed. There is just one possible predecessor state to completed which is the state activated. completed Subplans In Asbru suplans are used to group plan steps in various temporal orders. (See Asbru 7.2 Reference Manual [1] page 84 f. ) In the following representations only plans in the states active and completed are shown. sequential The plans are performed in strict order. Each is started after its predecessor is finished. This grouping type is displayed by bars that are displayed side by side in a row: Author: Wolfgang Aigner, 9755342 Page: 2 / 11

parallel All steps or subplans are started at the same time. They may finish at any time. Parallel plans are displayed by bars one at the top of each other with the same starting point (x coordinate of the beginning of the bar): any-order The subplans are performed one at a time, without strict ordering. E.g., the second step can be performed before the first one, if appropriate, but at each instant in time, only one step is performed, i.e., their execution must not overlap. As soon as one subplan is suspended, another one is started. If the suspended plan is reactivated, it is resumed only after the plan started meanwhile is finished. This type of plan grouping uses the same sort of display like the sequential plans. Æ see sequential unordered There is no ordering between the steps. They may overlap or not, and each may start and end whenever appropriate. The visualization of this grouping type uses a mix between the display types sequential respectively any-order and parallel: Repetitive plans Asbru offers the possibility to create various repetitive plans (cyclic plan, for-each plan and iterative plan) where the plan body is repeated as specified. Cyclic Plan Cyclic plans are displayed by the repetition of the drawn plan bar with the same color and caption. (For more details about Cyclic Plans see Asbru 7.2 Reference Manual [1] page 109 ff.) Author: Wolfgang Aigner, 9755342 Page: 3 / 11

For-Each Plan The body of a For-Each Plan contains a number of plans that can be executed in following temporal orders: parallel, unordered or any-order. This means that they are displayed as described above in the Subplans section. (For more details about For-Each Plans see Asbru 7.2 Reference Manual [1] page 114 ff.) Iterative Plan Simple sequentially executed loops are in Asbru modelled by an iterative plan. The display of the loop body is done in the same way as described in the Subplans section. (For more details about Iterative Plans see Asbru 7.2 Reference Manual [1] page 117 ff.) Caption The bars are visually assigned to a plan by their caption. Therefore the title attribute of the element plan is used. If this attribute is not specified, the attribute name is displayed which is required in the plan definition. Use of color Most colors do have certain meanings in our culture. i.e. red means attention, caution, hot, stop; blue means cold; green means o.k., go on, For this reason the use of color for plan bars is misleading and not useful. Plans do not have a semantic meaning per se, they are just a container for certain actions or processes. But it is necessary to have a way to distinguish between different plans and recognize if one plan is processed several times. Therefore we decided to use grayscale for the display of plans. Another possible solution would be the use of one color with changing the saturation for various plans. But as human perception is much better on distinguishing various shades of gray, we decided to use grayscale. Granularity The structure of a plan is tree like. That means that each plan consists of a plan body where other plans are executed. These plans can contain other plans and so on. Simple actions are the leafs of this tree. When displaying all leafs of the executed plan, the screen would be cluttered and full of very small sections where you couldn t recognize anything. Therefore we reduce granularity by only displaying at most seven plan bars at once. In the special case that seven plans are in parallel and there is not enough vertical available, fewer parallel bars are shown with an additional sign for indicating that there are more plans which can be accessed by zooming in. Expanded and collapsed view Because vertical space is very valuable on the screen, a plan can be displayed in expanded and collapsed view. Expanded view means display like described above, thus bars with caption. Using the mouse one can change the view to collapsed mode, which means that all bars shrink vertically to two pixels of height and the caption disappears. This view consumes very little vertical space while preserving the overview of plans. Expanded view: Author: Wolfgang Aigner, 9755342 Page: 4 / 11

Collapsed view: Displaying the future A further requirement for this tool is to show the course of plans in the future. This can be done by the element typical-duration of each plan. Plan selection takes places right after the predecessor plan stopped (preselection phase). That means that no exact course of plans can be carried out for the future. Because of the various possibilities future plans can be selected or not, future plans can only be displayed at a very global level due to our granularity restrictions. Plans in the future are displayed as a non-filled bar with dotted edges: Fisheye view For viewing parts of a plan at a very detailed level without losing the orientation in the whole plan fisheye view is very practical. That means that the area of interest is displayed at the decided precision and to the left and right of that section a compressed, low detail display of the surrounding is shown. Example: compressed, low detail focus of interest compressed, low detail Possible plans Possible plans are displayed as a separate LifeLine in parallel to the really executed ones. Plans are in the state possible if they can be executed due to certain parameters and restrictions. The plan execution unit selects one of the various possible plans for execution. But also each other plan that was in the state possible could be selected at that point. Therefore it is valuable information to see which other plans could have been selected also. Due to the fact that the display focuses on executed plans, possible plans are displayed in collapsed mode as default. Author: Wolfgang Aigner, 9755342 Page: 5 / 11

2. Visualization of values In principle Asbru deals with three different types of values: parameters, variables and constants. As you will see, only parameters can be displayed time dependent in parallel to the plan display. Types / Time dependence Parameters Parameters are time-stamped values and the Asgaard framework explicitly models their changes over time. The Monitoring Unit stores their state at any point of time in the past. Parameters have a certain value range within they are valid. When values out of that range are entered, they are marked as invalid. There can also be a certain interval of time defined within a parameter remains valid. That means that an entered value remains valid as long as no new value is entered or the time interval runs out. Thereafter the parameter is marked as invalid. Variables The second type of values in Asbru is called variable. This type is not time-stamped and is therefore only for the current instant valid. They are not stored and no value of any point in the past can be accessed. Plan arguments do have the same properties as variables. Constants The third basic value type is called constants. Constants are fixed at their declaration and remain the same during the plan execution. Examples for constants are date of birth or sex. Qualitative, time dependent display If a parameter refers to a qualitative scale, it can also be displayed via LifeLines due to the mentioned advantages. In addition to an intuitive color mapping this form leads to a very simple display that is easy to understand. A parameter displayed by its qualitative scale is at its simplest form a LifeLine with color sections representing a certain value of the scale. If a parameter gets invalid the bar is interrupted as long as the value remains invalid. Use of color, color mapping The most problematic issue of this display type is the color mapping. Because of the fact that most colors do have a certain meaning in our culture no automatic mapping can be performed without certain semantic knowledge about the scale entries. Asbru offers no possibility to add semantics in the definition of qualitative scales. Therefore we assume that there is a certain ontology that maps scale entries to colors. In addition it is possible to change this mapping due to the users needs. Example: <qualitative-scale-def name="fever"> <qualitative-entry entry="no-fever"/> <qualitative-entry entry="moderate-fever"/> <qualitative-entry entry="high-fever"/> </qualitative-scale-def> color mapping: Author: Wolfgang Aigner, 9755342 Page: 6 / 11

scale entry no-fever moderate-fever high-fever color LifeLine: Quantitative, time dependent display Quantitative values are displayed as diagrams where values at a certain point of time are displayed as dots or other bullets. These data points can be connected via straight lines or even other methods like interpolation can be used. The display mode should be user driven and the default display provides just the data points. Displaying quantitative time dependent values is a problem due to very small useable vertical space. This problem can be solved by providing the possibility of resizing the diagram vertically or viewing just parts of the vertical scale. But all these possibilities should be user driven and not automatically be done. The default view should just give a feeling for the overall change of the value and not be used for exact data analysis. One advantage of the quantitative display is the possibility of showing several parameters within the same diagram area. Therefore one can view several parameters at once and gets information about certain relations between them. The specially marked data points (in this case the red ones) identify the currently valid or numerically displayed values. These values are displayed numerically in the panel for displaying variables and constants as described below. Default display, switch between modes If a parameter has both a quantitative and a qualitative scale, the qualitative one is displayed as default. Switching between modes is done by the mouse. Author: Wolfgang Aigner, 9755342 Page: 7 / 11

Expanded and collapsed view Qualitative time dependent displayed values can be displayed in collapsed view as well. Qualitative values are shown in collapsed view as default. That means that the bar is shrunk to two pixels height as it is done in the collapsed view of plans: When quantitative, time dependent values are displayed in collapsed view, only the points in time where measurements were taken are shown. This view doesn t show anything about the value of a parameter but just shows when measurements were taken. Granularity A special problem of displaying time dependent parameters is their frequency in relation to the chosen time window. I.e. if you display very high frequency data in a time window of days or even month not every measured value can be displayed. Therefore we use the median function for displaying such parameters. Fisheye view As already mentioned in the Visualization of plans section, fisheye view leads to a detailed display at the focus of attention without losing orientation in the overall picture. For the display of quantitative values this means that values at the focus of attention are shown at decided precision and to the left and right of that focus the time scale is shrunk and only a few values with less granularity are shown. As qualitative values are displayed as LifeLines, this leads to the same result as the fisheye view of plans. compressed, low detail focus of interest Which values are displayed? Due to the vertical space limitations only very few time-dependent values can be displayed. We decided to display at most three parameters at once in collapsed qualitative mode or in one diagram panel in quantitative mode. Further we assume that there is a certain external ontology providing the program with the default displayed values for a given plan. Variables and constants As variables and constants are not time-stamped and no values of the past are stored only the currently valid values can be displayed. These values are displayed in a separate panel at the right side of the assigned LifeLines of one patient. The values are displayed as text and even to qualitative values no color mapping is provided. Author: Wolfgang Aigner, 9755342 Page: 8 / 11

3. Time cursor For selecting a certain point in time of interest a so-called time cursor is introduced. This time cursor is simply a vertical line that can be dragged per mouse. The exact position of the time cursor is shown numerically at the top and at the bottom of it. This time cursor triggers i.e. the currently numerically displayed values of a parameter (the marked data points of a diagram). exact value time cursor 4. Time scale One time scale that determines the displayed granularity is provided for all parallel shown LifeLines and diagrams. That is all displayed items are synchronized. This time scale is represented visually at the top and at the bottom of the screen. Fisheye view As described above fisheye view is used and therefore the time scale has no linear granularity. Vertical lines at the beginning and at the end and a white background display the area of focus. The border areas are shaded with a light color to provide the best contrast for the focus of attention. These areas are also draggable by mouse and so the focus of attention can be resized and moved. compressed, low detail focus of interest compressed, low detail Point of reference There are two possible time scales for the display: Plans and values assigned to real date and time In this case the time scale represents real points of time the point of reference is the earliest beginning of a displayed plan. Therefore two identical plans applied to two different patients and started at different points of time are shown shifted. One can compare states at certain points in real time but not values at certain plan states. Author: Wolfgang Aigner, 9755342 Page: 9 / 11

Plans and values assigned to the duration of plans Here the time scale represents points in the duration of plans the point of reference is the beginning of all plans. In this case one can compare plans and values at a certain point of plan execution but not states at a certain point of real date and time. 5. Panels and their order Parameters and plans are displayed one on the top of each other having a panel handle at the left. This handle is a rectangle containing a triangle which indicates whether the display mode is expanded or collapsed. The mode can be changed by clicking this triangle. Further it should also be possible to reorder the panels i.e. to display one qualitative value of all patients atop of each other for comparison. For the reason of easy assignment all panels of one patient do have the same background color as the variables and constant panel to the right. 6. Putting it all together Basically the screen is divided into three main areas: Time scale This area is split up into two panels one at the top and one at the bottom of the screen. It contains the time scale with its caption. Plans & diagrams That is the main part of the display, which contains the LifeLines and diagrams for the displayed patients. Variables, constants and current values This panel is located to the right of the Plans & diagrams area. It contains information about the patient (i.e. name, sex), certain variables and the numerically displayed parameter values. Author: Wolfgang Aigner, 9755342 Page: 10 / 11

7. Example This example is taken from the Asbru protocol for the Management of Hyperbilirubinemia in the Healthy Term New-born [4]. It shows the full screen of this tool. 8. References [1] A. Seyfang, R. Kosara, and S. Miksch, Asbru 7.2 reference manual, Technical Report Asgaard-TR-2000-3, Vienna University of Technology, Institute of Software Technology, (2000). available at http://www.ifs.tuwien.ac.at/asgaard/asbru/asbru 7 2 new/. [2] Catherine Plaisant and Anne Rose. Exploring LifeLines to Visualize Patient Records. University of Maryland. Human-Computer Interaction Laboratory. A.V. Williams Bldg. College Park MD 20782, http://www/cs.umd.edu/projects/hcil, <plaisant@cs.umd.edu>, <rose@cs.umd.edu>, Technical Reports CS-TR-3620 (1997). [3] PhD. Catherine Plaisant, PhD. Richard Mushlin, MD. Aaron Snyder, Jia Li, Dan Heller, and PhD. Ben Shneiderman. LifeLines: Using Visualization to Enhance Navigation and Analysis of Patient Records. University of Maryland. Human-Computer Interaction Laboratory. A.V. Williams Bldg. College Park MD 20782,. IBM T.J. Watson Research Center. Kaiser Permanente Colorado. In Proceedings of the 1998 American Medical Informatic Association Annual Fall Symposium, pages 76-80, 1998. [4] Asbru protocol for the Management of Hyperbilirubinemia in the Healthy Term Newborn, 9.1.2001 [5] Andreas Seyfang, Silvia Miksch, Mar Marcos, Combining Diagnosis and Treatment using Asbru, Vienna University of Technology, Institute of Software Technology, MedInfo 2001. Author: Wolfgang Aigner, 9755342 Page: 11 / 11

SQUEAK Prototype Guideline Overview Tool: SQUEAK Prototype 1. Introduction To demonstrate the ideas of our Guideline Overview Tool we developed a prototype using the programming language SQUEAK 1. We used the Version 3.0 which is the first version entirely based on the Morphic UI 2. 1 SQUEAK is an open, highly-portable Smalltalk-80 implementation whose virtual machine is written entirely in Smalltalk, making it easy to debug, analyze and change. Squeak comes under an open source license, meaning that you can download and use it for free. For further information see: http://squeak.org 2 SQUEAK switched from the Smalltalk MVC model to a new UI model called Morphic. Morphic is a user interface framework that makes it easy and fun to build lively interactive user interfaces. Morphic handles most of the drudgery of display updating, event dispatching, drag and drop, animation, and automatic layout, thus freeing the programmer focus on design instead of mechanics. Some user interface frameworks require a lot of boilerplate code to do even simple things. In morphic, a little code goes a long way, and there is hardly any wasted effort. [1] Author: Wolfgang Aigner, 9755342 Page: 1 / 4

SQUEAK Prototype 2. Prototype features The actual prototype only supports a subpart of the suggested UI in Guideline Overview Tool:. All in all the existing prototype gives a good impression of the intentions of the overview tool. Included features - display of all kinds of plan states (activated, suspended, aborted, completed) - display of all possible temporal orders (sequential, parallel, any-order, unordered, repetitive plans) - interactive switching between collapsed and expanded view - display of future plans - display of possible plans - display of qualitative values - free configurable bar colors - task, bar and time line captions - fully configurable time line (start and end point, granularity) Not yet included features - fisheye view - display of constants - visualization of quantitative values - time cursor - zooming, interactive changeable time scale - changing point of reference 3. Testing the prototype The prototype can be started within the Squeak programming environment or within a browser by installing the Squeak browser plug-in. The enclosed CD-ROM includes the full Squeak image and the file LifeLine.st. When there is already a Squeak programming environment installed, just this file has to be filed in. The test project is started by invoking the method testproject of the class TestLifeLine. 4. Structure of UI parts The structure of the prototype is a tree of morphic elements as follows: TimeLine GroupingFrame GroupingFrameCollapseHandle CollapseTriangle LifeLineGroup LifeLine root element group elements of 1 st LifeLine (bilirubin) collapse / expand handle for LifeLine triangle of handle group of LifeLines qualitative value bilirubin yellow part green part caption bilirubin GroupingFrame group elements of treatment plan for patient 3 GroupingFrameCollapseHandle collapse / expand handle for LifeLine CollapseTriangle triangle of handle LifeLineGroup group of LifeLines LifeLine treatment plan for patient 3 / part 1 Author: Wolfgang Aigner, 9755342 Page: 2 / 4

SQUEAK Prototype task ask age caption Ask age task ask age caption Ask age LifeLine treatment plan for patient 3 / part 2 task task task diagnosis caption Diag task feeding caption Feeding LifeLine treatment plan for patient 3 / part 3 task observation caption Observation task phototherapy caption Phototherapy caption Patient 3 GroupingFrame group elements of possible plan for patient 3 GroupingFrameCollapseHandle collapse / expand handle for LifeLine CollapseTriangle triangle of handle LifeLineGroup group of LifeLines LifeLine treatment plan for patient 3 / part 1 task ask age caption Ask age task ask age caption Ask age LifeLine treatment plan for patient 3 / part 2 task task Aborted aborted task diagnosis caption Diag task feeding caption Feeding LifeLine treatment plan for patient 3 / part 3 task observation caption Observation Suspended suspended task phototherapy caption Phototherapy caption Patient 3 Patients 2 & 1 GroupingFrame group elements of time line ruler GroupingFrameHandle nonvisible element for spacing TimeLineRuler timeline AlignmentMorph start point caption caption 1/23/2000 caption 09:00:00 AlignmentMorph spacer AlignmentMorph end point caption caption 1/24/2000 caption 06:00:00 Author: Wolfgang Aigner, 9755342 Page: 3 / 4

SQUEAK Prototype 5. References [1] John Maloney, An Introduction To Morphic: The Squeak User Interface Framework, Walt Disney Imagineering, 2000. available at http://coweb.cc.gatech.edu:8888/squeakbook/uploads/morphic.final.pdf. Author: Wolfgang Aigner, 9755342 Page: 4 / 4