Home

Archie - Software Composition Group

1. return Status CANCEL STATUS Do actual work Fill result array 70 monitor worked 1 set content this dataSet setContent result job finished successfully monitor done return Status OK_STATUS Listing C 3 shows a skeleton of an actual createContent method The first thing you might notice is that the method does not return the actual data but the status of the job If the job is finished the method should return an OK status It is also important to check for cancel request inside loops if the method doesn t check for that the cancel button in the sidebar will not work You should also set the size of your job and work with the monitor so users get feedback on how your job is progressing When the job is finished be sure to set the content of the dataset with this dataSet setContent result else it will not work The most important part is that you put the right data into the dataset As you can see in the example you should fill a list which contains arrays of objects that implement the comparable interface If the objects you want to put into the dataset don t implement said interface you might have to write a wrapper class for them The reason is that we have to make sure that the results are sortable in the result view The arrays in the list will end up as rows in the result table they have a determined size which has to be the same as the headings size The list holds all these
2. Weir09a Weir09b John Arthorne Chris Laffra and Gary Johnston What is a plug in fragment http wiki eclipse org FAO What is a plug in fragment 3F February 2008 Donald Bell UML s Sequence Diagram http www ibm com developerworks rational library 3101 html February 2004 Heinz Bhend TEXTDOKUMENT der CD ROM Die elektro nische Krankengeschichte in der Arztpraxis DOCUMENT to CD ROM October 2006 A L Cochrane and Max Blythe One man s medicine Memoir Club 1989 David Gallardo Developing Eclipse plug ins http www ibm com developerworks opensource library os ecplug December 2002 Bundesamt f r Gesundheit Strategie eHealth Schweiz http www ehealth admin ch June 2007 Chris Laffra and Nick Veys What is an extension point schema http wiki eclipse org FAO What is an extension point schemaf3F June 2006 Jeff McAffer and Jean Michel Lemieux Eclipse Richt Client Plat form Addison Wesley 2007 Michael Valenta On the Job The Eclipse 3 0 Jobs API http www eclipse org articles Article Concurrency jobs api html September 2004 Gerry Weirich elexis buchhaltung basis http www elexis ch jp content view 236 75 January 2009 Gerry Weirich Verbreitung von Elexis http www elexis ch jp content view 245 87 February 2009 77
3. 3 Heute Bp pai E Konsu GS Labor Rezep AUF A Konsultatio 53 Konsultation 9 E SE 8 Vorname ce Aapxkluo Uhmeoencb 1795 ansan Bezahlt nzi Kontakte e PAN aapxkluo Uhmeoench Dk A berd Geschlecht Mobil Resultate besprochen alles A Acowegn Pxkowhy w 08 01 1 jio bis auf rez A Acpjsw Ukwxhnm w 06 08 19 Kopfschmerzen Lufu normal Briefe D adxsfmF Oquemml 11 09 15 tomang Telefon 1 SS ep goce A afolyn Obbwxiti m 25 06 198 normal 29 11 2007 a Afwgaogdje Ueebfegffge m 0 f Rezept A Agitighpwa Czglimdxe m 20 0 EVEN ERE Emal 23 11 2007 Bezahlt Leistungen A Agxkiqw Uwycgsihss m 04 03 555 6437414 Henzi A Ahdecuvfl Isledfm m 07 07 15 A Ahsnii Oeomkoofw 10 06 1924 Gruppe Konto A aibbwbd Kwrhhii m 31 12 191 Hatte Episode mit antike A Alemagsbn Noxntdatw 22 11 lWechen recesseti Jeet A Aifmrljw Qywjsoeldt w 01 02 I i N N 2TageBesser dazu amp Ainsenv Okhvdgbx m 19 04 1 Anschrift Frau Uhmeoencb Aapxkluo Bern Schwindel morgens keine iofi i belkeit kei B Aiofivw Beppisprvdtw 28 04 1 Zusatzadressen Sehst r ungen War nur 49 Reminder Gate Coen er gin 4 19 gt Diagnosen kg m kw k jkiuwx Miwbpgbmhm w 29 0 a Atembeschwerden mit D albescbvtip Wbozojlac w 08 044 gt PersAfnliche Anamnese Druck auf Brust va PS n gt Allergien morgens gel bei lt i gt g Anstrengung gel ege b gt Risiken Einschla
4. Override run setExistingViewDirty getView NewStatisticsAction ExportAction view StatisticsView view StatisticsView parameters ParametersPanel DEFAULT_EXTENSION String providerInformation ProviderInformatioPanel SINGLE SPACED FILES boolean listeners ArrayList lt IPropertyChangeListener gt ExportAction JOB RUNNING String ExportAction JOB DONE String Override run NewStatisticsAction saveFile Override run getNameSuggestion done Override addPropertyChangeListener Override removePropertyChangelListener update updateListeners aboutToRun awake running scheduled sleeping Figure 4 6 UML overview of the action classes 33 4 3 1 New Statistics Action This action is the main controller defined in the class NewStatisticsAction which is why it is probably the most important one It not only controls when a provider is scheduled and tells it to start gathering the data but also acts as a mediator between the result view main view and the sidebar view The sidebar view contains custom widgets necessary for setting parameters for a currently selected provider Widgets are described in Section 4 8 in more detail Each action object needs to extend the Eclipse API class Action The heart of each action is the run method This method is called when the us
5. ch unibe iam scg archie dataprovider gt lt DataProvider class ch unibe iam scg archie samples UserOverview name System User Overview gt lt DataProvider gt lt category id ch unibe iam scg archie dataprovider ProviderCategory name Samples lt category gt lt fragment gt A simple example of the user statistics is shown in Listing 4 2 where UserOverview needs to be a subclass of AbstractDataProvider described in the following sec tions The definition can optionally include a category element as described in Section 4 1 5 25 4 2 Data Providers and Datasets The heart of Archie is the abstract class AbstractDataProvider Plug ins that hook into Archie have to extend this class in order to provide the data they collect Furthermore an AbstractDataProvider extends the Eclipse API class Job contains a Dataset object as well as label and content providers for Eclipse viewers whose details are described in Section 4 2 1 Hans Friedrich 1 239 Figure 4 3 An Example Dataset Datasets are tables modeled as Java objects A dataset consists of a list of headings the top most row of a table and a content being the rest of the rows in that table In addition to this a dataset implements the Iterable lt Comparable lt gt gt and Cloneable interfaces which on the one hand allows us to retrieve an Iterator for a datasets content on the other hand to create duplicates of datasets for further
6. Another challenge is to find a solution to enable plug ins that use the frame work to be generically parametrizable This means a plug in can have an unknown number of attributes with unknown types and names and the framework must be able to present these attributes as editable param eters in the correct form to the user It also must know how to get the information from the user interface back to the plug in Each plug in needs to be able to programmatically define validation pro cedures for its parameters without having to concern itself with the user interface Moreover the user interface has to be able to show all available statistics to the user and help him in finding what he is looking for 1 3 EMR Systems and Elexis In the following paragraphs we shortly explain the context in which Elexis and thus also Archie is situated Most physicians still use paper and file based systems to manage their pa tients Adoption of Electronic Medical Record Systems EMR is slow Switzerland is still one of the least advanced countries when it comes to distri bution of EMR There are various reasons for the slow adoption at which we take a closer look in Chapter 2 Nevertheless the number of deployed EMR systems is slowly rising and there are various efforts being made to speed up the process Most notable is Switzerland s ambitious eHealth strategy which wants to break ground for a wide deployment of EMR its goals are to define
7. Consultations Total T Income IV Spending W Profit You can create a 3d chart by selecting this button below Note that the chart is not rendered by a 3d engine which can lead to distortions depending on the size of the container Y 3D Graph lt Back Finish Cancel Figure B 5 Chart Wizard Bar Chart Page BEE Choose Rows Please specify which rows you would like to include in your chart Date Consultations Total Income Spendin Profit 2008 01 593 CHF 39 520 80 CHF 2 857 27 CHF 36 663 53 2008 02 503 CHF 35 562 07 CHF 2 807 48 CHF 32 754 59 2008 03 514 CHF 37 019 00 CHF 3 452 16 CHF 33 566 84 2008 04 413 CHF 28 636 75 CHF 2 422 23 CHF 26 214 52 2008 05 2 CHF 80 69 CHF 4 43 CHF 76 26 lt Back Next gt o Cancel Figure B 6 Chart Wizard Choose Rows 66 Income Spendings per Month 2008 03 Category m Profit M Spending Figure B 7 Example chart created with Chart Wizard 67 Appendix C Developer Manual Writing a statistic which uses the Archie framework is pretty easy There are two main steps involved Extending the AbstractDataProvider class and Registering the statistic with the Archie Extension Point We will look into these steps in detail in the next sections For examples on how to write statistics please consult the source code of the Archie samples project You can find information on how to check out the source files via SVN from the Archie project websit
8. End Date 31 12 2008 Cohort Size 5 Active Mandator Only true Consultations Total Costs Average Costs Total Profits CHF 41 31 CHF 5 90 CHF 732 71 CHF 75 30 CHF 1 30 CHF 2 694 06 CHF 46 45 CHF 396 90 CHF 3 71 CHF 6 055 33 CHF 56 59 CHF 1 001 28 CHF 5 66 CHF 10 781 14 CHF 60 91 CHF 1 670 53 CHF 7 05 CHF 15 577 81 CHF 65 73 CHF 1 341 62 CHF 4 61 CHF 19 269 74 CHF 66 22 CHF 838 11 CHF 4 37 CHF 12 665 12 CHF 65 96 CHF 786 03 CHF 5 82 CHF 9 032 57 CHF 66 91 CHF 722 31 CHF 6 28 CHF 8 075 55 CHF 70 22 CHF 729 56 CHF 5 49 CHF 9 175 92 CHF 65 99 CHF 792 38 CHF 7 14 CHF 7 974 70 CHF 71 84 CHF 642 59 CHF 6 76 CHF 5 894 28 CHF 62 05 CHF 488 27 CHF 6 69 CHF 4 813 72 CHF 65 94 CHF 605 04 CHF 6 05 CHF 6 125 63 CHF 61 26 CHF 642 03 CHF 7 55 CHF 4 503 44 CHF 52 98 CHF 461 06 CHF 8 54 CHF 2 985 35 CHF 55 28 CHF 250 32 CHF 4 91 CHF 2 690 02 CHF 52 75 CHF 58 93 CHF 14 73 CHF 228 65 CHF 57 16 Figure B 3 The Output View The Chart Wizard The Chart Wizard allows you to create charts from any result table To explain the Chart Wizard we show an example creation of a chart from a result set When you click on the action button you are presented two options Creating a pie chart or creating a bar chart which can also create line charts as seen in Figure B 4 The next dialog asks you to specify what exactly you want to plot and how as can be seen in Figure B 5 When all parameters are set you can choose which rows of the dataset you wan
9. Tables wrapped in a TableViewer can thus be populated with any type of objects the content provider of a viewer understands In order to allow our TableFactory to generate results and create tables from any kind of data an AbstractDataSource has a DataContentProvider and a DataLabelProvider associated with it Both providers are subclasses of provider objects from the Eclipse API This also allows different data sources to declare their own providers and control how the data is being displayed to the user Figure 3 4 shows these architectural and design decisions in more detail In addition to these model based decisions we decided to create a dual pane user interface for Archie This means to have a settings pane on the right hand side where users can chose what statistics they want to create and adjust their parameters and show the results from these statistics in a pane on the left side During the development of Archie the architectural and design decisions de scribed in previous section were constantly refined The main idea of having some sort of central data provider modelled as an Eclipse Job remains Also the integration into the Eclipse API via Content and LabelProvider is still the same in the final version Some implementation details are missing in the initial design in Figure 3 4 especially questions such as How exactly do the data providers store raw data from the database for further processing How does do the chart
10. 3 The Chart Model The underlying object in a chart wizard is the chart model The chart model is a simple class that contains definitions and settings about how a chart based on this model should be rendered what kind of chart it is and what data from a dataset the chart contains 40 Listing 4 5 Chart Model Object Variables public class ChartModel Listing 4 5 shows the object variables in a chart model jae x Constant for pie chart types 1 public static final int CHART_PIE 1 fax x Constant for bar chart types 2 Bar charts can also be handled as line x charts as they both are created from a category dataset There s a switch x in the bar chart type that can be activated for line charts public static final int CHART_BAR 2 This switch can be activated for bar charts which makes them render as line charts private boolean isLineChart private String chartName private DataSet dataSet private int rows private int columns private int keyslndex private int valueslndex private int categoryColumnlndex used in bar amp line charts private int chart Type private boolean is ThreeDimensional are set in the chart wizard steps the chart model method isValid returns whether a chart model is valid and can be used for chart creation Each chart model has a chartType which is either ChartModel CHART_PIE or ChartModel CHART_BAR Although there are three types
11. Date co Y E Calendar Field End Date 31 12 2009 with Datepicker Cohort Size 5 2 v o Mo Di Mi Do Fr Sa 1 23 45 6 7 8 9 1011 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 Active Mandator Only _ Y 4 Abfrage Starten amp Abbrechen 44 Dez 2009 rr Datum Ausw hlen ee Checkbox Field Figure 4 5 Three different UI widgets Archie provides four different widget types that can be used together with annotations a simple text widget a numeric widget a date widget and a checkbox A screenshot showing all widgets is depicted on Figure 4 5 Bach of these UI widgets provides default validation and quick fix methods A date widget for example is rendered together with a date picker pop up for users to be able to chose a date quickly and easily Moreover additional val 31 idations using regular expression patterns can be set by each data provider implementation using annotations 4 2 3 Provider Implementations Implementing a custom data provider is a two step process First the im plementing plug in needs to hook into the data provider extension point as described in Section 4 1 3 As stated in the extension point definition a plug in must have at least one but can have as many as wanted implementation of the AbstractDataProvider class The second step consists of defining an implementation of a data provider This is where the plug ins do the statistical computations and wr
12. computer we had to separate the coding areas This way not both of us were working on the same task simultaneously and thus could advance faster while still having the possibility to discuss problem areas as well as possible solutions together This process proved to be very effective Before we started working on the report we defined the initial table of con tents together based on which we assigned specific chapters to each of us We wrote the chapters individually and apart from each other before we cor rected and combined them as a pair to one entire document in an iterative process Dennis wrote initial versions of Chapter 2 Appendix B Appendix C and Chapter 6 Peter wrote initial versions of Chapter 1 Chapter 3 Chap ter 4 and Chapter 5 We did not assign the chapters according to a specific scheme For the task management process we made heavy use of activeCollab a Inttp www activecollab com 52 project management and collaboration tool as well as Subversion for ver sioning source code and text files Near the end of the project we defined a Google Code project website for Archie in order to maintain official releases and downloads as well as provide a public interface for reporting issues and enhancement requests nttp subversion tigris org 3http code google com p archie 93 Chapter 6 Requirements Engineering and Validation At the beginning of our project we visited various doctors and
13. down list of available statistics in the Archie perspective prefixed with the name of the category they are in C 4 Where to Get Additional Help You can find help for Elexis development in general at the Elexis forum at http www elexis forum ch For help with Archie development you can go to the official project website at http archie designchuchi ch or the already mentioned Google code website at http code google com p archie 74 List of Figures 2 1 2 2 3 1 32 3 3 3 4 4 1 4 2 4 3 4 4 4 5 4 6 4 7 4 8 4 9 4 10 4 11 4 12 4 13 4 14 4 15 4 16 fel BA 12 B 3 B 4 B 5 B 6 B 7 Schematic concept of a sample EMR 9 EMR EHR CIS in sample relation 10 Sanclipse Prototype Overview 14 Eclipse Plug in Architecture 15 Elexis Architecture lt lt is wu Yale ee a wer sah ee a 16 First Conceptual UML Draft 17 Extension Schema Definition Screenshot 23 List of Providers Before and After the Extension Point Changes 24 An Example Dataset x sx 4460 Sk me ea ea 26 AbstractDataProvider class and surroundings 28 Three different UI widgets 2 2 22 220 0 31 UML overview of the action classes 33 Data Provider Controller Interaction 35 UML overview of the manager classes 37 UML overview of the factory classes 2 222 220 38 Dataset Mapping in Bar Charts cc 40 V
14. http www gnu org licenses lgpl html 3http www openoffice org product calc html http office microsoft com excel 20 Chapter 4 Implementation 4 1 Architecture 4 1 1 Structure of an Eclipse Plug in As described in Section 3 2 we decided to create Archie as an extension of Eclipse and Elexis A plug in the basic building block of Eclipse contains a collection of files and a manifest file MANIFEST MF describing the rela tion of the plug in to other programs packages and libraries The collection of files can include program source code read only content such as images translation message files used for internationalization documentation or oth ers In addition to the manifest file plug ins also have a plugin xml file In former Eclipse versions particularly 3 0 and older the plugin xml used to contain execution related information but since Eclipse version 3 1 this has been moved to the MANIFEST MF file The plugin xml however still contains im portant information namely the definitions of extension points the plug in either uses to extend other parts of an application and or extension points it defines for other plug ins to extend McAf07 4 1 2 Archie Plug in Definition Archie consists of two parts a main plug in and a plug in fragment A plug in fragment is considered as an optional part of a plug in The main difference between a plug in and a plug in fragment is that a fragment needs 21 to define a F
15. lt interface gt gt Eclipse API N IContentProvider fctipse aer P e IN Eeli API J lt lt interface gt gt LabelProvider ec P BackgroundJob IStructuredContentProvider i Extension Point 258 DataContentProvider DataLahelProvider abstract DataSource Annotations public IContentProvider contentProvider Goiter amp Setter public LabelProvider labelProvider 7 gt public int DataSet public IContentProvider getContentProviden public ILabelProvider getLabelProvided public DataSet getDataSet BackgroundJob Listener 1 Li 1 1 eh A N Li x SomePlugin Y SomePluginDataSource 1 1 ChartFactory TableFactrory Figure 3 4 First Conceptual UML Draft Furthermore we need two controllers for creating tables and charts Tables are used in the user interface UI to display the gathered data to the user Charts are used to build visualizations of data already displayed in the UI For this we define two controller classes a TableFactory for generating tables and a ChartFactory for creating pie line or bar charts based on the data in AbstractDataSource 17 Tables in Eclipse can be wrapped in a TableViewer object Content of a table in Eclipse can only be built using strings whereas viewers can have a content and label provider These providers define whether and how elements in table rows are being displayed
16. net 10 However these issues are being worked on and awareness for the issues is growing Solutions are actively worked on internationally and nationally In Switzerland as noted above the government started its eHealth initiative of which one major subproject is to find and specify standards The current lack of standards many times also the non compliance to avail able standards and the resulting lack of interoperability between existing software systems are some of the reasons for the slow Adoption of EMR and other health information technology In spite of studies showing revenue gains after implementation of EMR the healthcare industry spends much less of its budget on IT than other infor mation intensive industries Aside from the immaturity of standards there are several industry specific issues responsible for this One reason is the difficulty to incorporate older records of patient data into the new systems The process is time consuming and expensive and should be done following precise standards to ensure that all information concerning the patient is available in the new system Information on a patient may exist in any number of formats sizes media types and qualities which further complicates accurate conversion If a practice has used paper based patient management for many years or even decades the costs of a switch to EMR are very high Another major issue is privacy Individual electronic records have to
17. of charts we only define two chart types and add the variable isLineChart for that case as a line chart is a special case of a bar chart both depending on the same dataset type JFreeChart dataset as described in Section 4 6 2 41 These variables In addition to this a chart model also specifies which columns and rows from a dataset also stored in the model will be used for categories keys and or values in the chart s dataset JFreeChart also provides methods for creating three dimensional charts However the underlying datasets are independent of this setting which is why we only specify a boolean isThreeDimensional for switching to a 3D chart 4 7 Helper Classes Several places in our program source code make use of small and almost identical pieces of code Because of this we define certain utility or helper classes These classes help us to keep the code clean and structured and allow us to reuse common shared functionalities and methods in different places All helper classes contain only static methods We also expose the helper classes in Archie to other plug ins by defining entries for each helper class in the plugin xml file Jr Dashboard 7 Ausgabe Ansicht N EN Current Statistics Start Date 01 03 2008 End Date 31 12 2008 Active Mandator Only true Save As patienten kosten umsatz und gewinn m Where 5 Desktop _ Cancel Save Figure 4 11 Valid File
18. offers custom defined extension and provides interfaces for other plug ins to extend Elexis itself Figure 3 3 illustrates the architectural situation with regard to Elexis and the Eclipse framework Elexis consists of a main plug in encapsulating core functionality that enables it to run The main plug in also defines the exten sion points for other plug ins contains definitions about how Elexis is to be deployed on different operating system and holds all the image resources used throughout the UI Moreover the main plug in provides ant build scripts for automated building as well as Latex and JavaDoc documentation of Elexis for both users and developers 15 Eclipse Framework Elexis Main ch elexis ch elexis views ch elexis wizards j Elexis Elexis Agenda Elexis Befunde Kassenbuch Elexis Figure 3 3 Elexis Architecture The extending plug ins are dependent on the main plug in and cannot be run without it only extending or adding new functions to the main plug in Based on the architecture of Eclipse and Elexis one of our main goals provid ing an application that is easily deployable into existing Elexis installations can be easily accomplished by creating a plug in of our own 3 3 Architectural and Design Decisions We decided that Archie is an Eclipse plug in and provides some sort of an extension point allowing other developers to easily hook into Archie and provide statistical data about Elexis The next
19. processing The headings are modeled as a list of strings where each element is a heading of one column in the table The content is a list of Comparable lt gt arrays Datasets provide methods for accessing and updating headings entire rows and or individual cells An example of a dataset is depicted in Figure 4 3 The dataset class was called AbstractDataSource in our initial design de scribed in Section 3 3 We renamed it during the project Moreover when we first modeled the datasets their content was composed out of Object arrays However we later discovered that this was not ideal when used in combination with sorting Eclipse tables that were composed out of datasets this is why we decided that every object in a dataset needs to implement Comparable interface 26 Listing 4 3 Abstract Data Provider Methods public abstract class AbstractDataProvider extends Job jaa x Returns the description for this data provider x return Returns the description for this data provider public abstract String getDescription x Creates headings for each column in the dataset object of this provider x return A list of strings List lt String gt containing the headings protected abstract List lt String gt createHeadings This method should do all the work necessary to populate the dataset s x content It s called in the job s execute method after some x initializations have been done x retur
20. provider could define selection listeners that would react to double blick events on a table row or cell ac cordingly Adding More Statistics The samples part of Archie contains seven dif ferent implementations of the framework providing seven different statis tics Although statistics developed for Archie by other parties Weir09a already exist additional statistics implementations could be provided in the future 58 Appendix A Licensing of Archie From the very beginning our stated goal has been to distribute Archie as part of the official Elexis release To accomplish this goal we release Archie under the Eclipse Public License EPL the same license that Elexis uses 1 0 The EPL has lower restrictions than other licences such as the GNU General Public License GPL For instance every software building upon or using software licenced under the GPL has to be released with the same licence while this is not necessary under EPL Archie will become part of the Elexis core with the next major release Inttp www eclipse org legal epl v10 html http www gnu org licenses gpl 3 0 txt 59 Appendix B User Manual B 1 Quick Start To install Elexis and Archie follow these steps 1 Download an Elexis 1 4 installation package for your platform from http www elexis ch jp index php option content amp task view amp id 57 and install it on your computer Download Archie from archie designchuchi ch or dire
21. rows and is of variable size C 2 Adding Additional Parameters To present the user with additional parameters that you want to use in the creation of your statistical data you can write getter and setter methods for these parameters and annotate them in the right way What follows is an example of such a method pair which is responsible of providing the user with a checkbox in the UI asking whether he wants to create the statistic only for the active mandator involving only his consultations or for all mandators 71 involving all consultations in the system Providing users with such an option has been established as good practice during development Listing C 4 Annotated method pair for additional statistic parameters JET return True if statistic should be created for current mandator only false else si OGetProperty name Active Mandator Only index 1 widgetType WidgetTypes BUTTON CHECKBOX description Compute statistics only for the current mandator If unchecked the statistic will be computed for all mandators public boolean getCurrentMandatorOnly return this currentMandatorOnly Jos Oparam currentMandatorOnly OSetProperty name Active Mandator Only public void setCurrentMandatorOnly final boolean currentMandatorOnly this currentMandatorOnly currentMandatorOnly You are now able to use this currentMandatorOnly in your code the UI has been set up automati
22. set parameters get rendered and validated We provide implementations for most common widget types A list of avail able widgets is shown below a more detailed view of the entire class hierarchy is depicted in Figure 4 13 TextWidget Is an implementation of an abstract widget containing a sim ple text input field In contrast to the abstract widget a TextWidget contains a SmartField object a wrapper class for a widget s control that provides label decorations for signalising the validity of a widget s value to the user NumericTextWidget Is a subclass of the TextWidget providing an im plementation of text widgets which may only contain numeric charac ters A numeric text widget also has a custom implementation of the SmartField class a SmartNumericField in order to provide proper vali dation methods for numbers This class also takes care of parsing the content of the text widget in order to return an int value DateTextWidget A widget based on the TextWidget which additionally provides a date picker popup displaying a calendar to the user in order to simplify the input of dates CheckboxWidget A simple checkbox widget implementation The control in this widget is aSWT Button the value returned is either true or false depending on the selected state of the button 44 AbstractWidget label Label control Control getLabel Label createLayout void getValue Object setValue value Object
23. standards looking at financial feasibility law foundations basic ed ucation and online service possibilities It also wants to start pilot projects and in general speed up deployment of EMR dramatically on a pretty tight schedule There is not much diversity in currently available EMR systems in Switzer land there are a few and most of these software solutions available are either expensive or require the physicians to have their own large information sys tem which most physicians don t consider an option as they don t have the necessary knowledge about deployment and or maintenance Also the costs that such systems impose are high Elexis is an alternative to proprietary EMR solutions it is based on the open source Eclipse Platform and can be used with various open source database systems such as MySQL or PostgreSQL Inttp www ehealth admin ch 2http www eclipse org 3http www mysql com http www postgresql org 1 4 Archie At the beginning we were searching for a name for our project We settled on Archie after Professor Archie Cochrane 1908 1988 Coch89 He was a British general practitioner and originator of evidence based medicine which is every form of medical treatment where patient oriented decisions are made specifically on grounds of proven effectiveness which in turn is given through the means of statistical data 1 5 Structure of this Document In Chapter 2 we give a short overview of the pr
24. step is to define an application design that allows us to meet the goals set in Section 1 2 1 Create a framework for developers allowing them to guickly and easily assemble their own statistics 2 Give users the possibility to guickly generate and export textual and graphical reports about any data in Elexis 3 Create accompanying statistics about important Elexis data 16 First we define an abstract class AbstractDataSource that provides a first entry point for other plug ins to extend and write their own statistics This AbstractDataSource extends the Elexis class BackgroundJob which is a sub class of Eclipse API class Job In Eclipse Jobs are tasks that can be scheduled and executed During their execution the API provides methods for monitoring locking resources or cancelling currently running jobs Eclipse Jobs also provide user feedback during execution Vale04 These properties are perfectly suited for our data providers Data providers are gathering statistical data about Elexis During this process a visual feedback about the current progress is shown to the user This is the the same concept Elexis already uses to retrieve data from the database most parts of Elexis itself where data retrieval takes place particularly if the retrieval may take some time use a BackgroundJob and therefore the Eclipse Job API lt lt interface gt gt ILabelProvider A Eclipse API by i Job Eclipse API lt
25. test user and click on apply B 2 Usage To use Archie open up the Archie perspective by either clicking on the button in the main button tool bar or by clicking on Window Open Perspective Other Archie Perspective in the menu bar In the perspective you see two views the Dashboard and the Sidebar B 2 1 Dashboard View The Dashboard gives a quick overview of your Elexis system It shows some key data like number of patients and number of consultations With a click on the Create Charts action button on the upper right side of the view the dashboard draws four charts Number Of Consultations Shows number of consultations during the last six months Consultation Money Shows how much profit income and costs we had for consultations during the last six months Costs of Consultations Shows costs of all consultations in a histogram grouped by age group and gender Age Histogram Shows all patients in the system grouped by age group and gender The size of the age groups of the last two charts can be altered in the Archie preferences To open them click on File Settings Archie 61 A ee E GE m Profit Spending Income SS m Male M Female m Male M Female Figure B 1 Dashboard View 62 B 2 2 The Sidebar View M Statistic Parameters Creates a list with all consultations in the given timespan displays the total and the average costs and the total and av
26. void setDescription description String void isValid boolean i ls CheckboxWidget TextWidget oe ComboWidget regexValidation RegexValidation controlDecoration ControlDecoration smartField SmartField getSelection boolean getValue Object isValid0 boolean setValue value Object void setDescription description String void createSmartField void getValue Object setValue value Object void isValid boolean handleModify smartField SmartField void hasRegexValidation boolean setDescription description String void showDecoration smartField SmartField type int show boolean void DateWidget NumericWidget setltem items String void getValue Object isValid0 boolean setValue value Object void setDescription description String void datePicker DateTime datePickerShell Shell popUpCalendar void popDownCalendar void updateCalendarPopup void createQuickFixMenuffield SmartDateField Menu createQuickFixMenuffield SmartNumericField Menu getValue Object Figure 4 13 Class hierarchy diagramm of UI Widgets 45 4 9 Elexis User Interface Contribution As previously stated one of our goals is to create a modern good looking and functional user interface Elexis itself in our opinion needs some overhauls in
27. we hope to find the ap propriate method of packaging and deploying Archie as part of Elexis This is one important criteria another one is that users should be able to easily install or uninstall Archie with all its functionality Fortunately every Eclipse application is modular composed out of small functional parts into one application Everything in Eclipse is a plug in Gall02 The term plug in in Eclipse refers to the unit of modularity Except 14 Platform runtime Your Plugin Figure 3 2 Eclipse Plug in Architecture for two indispensable core plug ins the Workspace and the Workbench an Eclipse based application can be completely stripped down and built up from the ground by using custom plug ins that extend the core functionality Plug ins in Eclipse provide extension points that can be used by other plug ins to add or change functionality of the providing plug ins The Workspace plug in allows other plug ins to extend the Eclipse user interface to con tribute menu items or entire menus and to create new views and add ad ditional buttons to the toolbars The Eclipse framework of course provides plug ins that already extend the core but can also further be extended by other plug ins An illustration showing the Eclipse plug in architecture is depicted in Figure 3 2 Elexis is an Eclipse RCP based application which means it extends Eclipse by implementing the extension points defined in the framework but also
28. 015 every citizen should have his own patient record Gesu07 How this goal should be accomplished is not drawn out clearly yet but the example of Australia shows that it is feasible There the percentage of doctors using EMR ascended from 8 in 2000 to over 50 in 2006 Bhen06 About 90 of practices in the Netherlands Great Britain and Scandinavia use EMR The quality of these systems varies strongly though Some nations advance implementation of EMR through subsidies and or regulations The USA are currently in a similar situation as Switzerland with a distribution of EMR of about 15 Bhen06 2 1 Terminology If one starts to engage himself with the domain of EMR he faces confusion and finds no general consensus on key aspects There are many abbreviations for different concepts in the world of electronic patient records and many have disputed conflicting or overlapping definitions Some of these problems occur when businesses try to sell their products to customers and their marketing departments create new buzzwords to stress uniqueness Another reason is that standardization efforts in the field are still in an early phase and sometimes uncoordinated which is reflected in the jungle of different words with different meanings In the following paragraphs we try to flesh out some of the core concepts we already used so far to make them clearer An Electronical Medical Record EMR is a repository of demographic and health data of a p
29. 50 important in medical practices where not only the physician might have ac cess to Elexis As Archie provides access to sensitive statistics about patients access to this data needs to be restricted in certain cases Unfortunately the entire access control protection only works when Elexis is started with the de CH language parameter and thus localized in German This problem is known to the Elexis core developers but at the point of this writing there is no working solution which is why the ACL restrictions do not work in Elexis language setting other than German This is a limitation of the currently stable Elexis 1 4 release and has been fixed in the latest 2 0 beta version ol Chapter 5 Team Organization Because of the uncommon situation of doing this project as well as writing this report in a team we needed to divide the tasks in each phase of the project The phases of the project included creation of the initial project presentation implementation of the prototype making architectural desi cions for the programming part programming Archie along with the sample statistics plug in and at last writing this thesis Throughout the project we tried to work in each phase in pair when it was efficient and divide the tasks when it was not We made the initial project decisions and discussed the entire architectural process together During the programming part although working side by side in the same room each on his own
30. 93 Gruppe Konto ghne v ein y lt Inon F lltext werden Ein S Lisa Mona w 14 03 1969 EE 0 00 F ltext zu werden Piel Mille Tonnere m 24 06 1972 erfordert viel d Muschter Hans m 02 04 1945 Einf hlungsverm gen in M schterli Fritz m 02 05 1983 Anschrift Herrn Arriba Samis Enge Gasse 17 CH 9999 Elexikon die Feyche oinor verwaisten allei Prefect Ford m 21 12 2018 EEN stehenden L cke Das Wu Louis m 02 01 1998 N Studium zum E Reminder gt Diagnosen anerkannten F lltext gt Pers nliche Anamnese umfasst 6 Silvester und la erfordert eine gt Allergien Funfundzwanzigstiindige isik Abschlusspr fung Ich Rechnungen F lle 3 x b Risiken m chte meinen Eltern gt Bemerkungen ea und Lorem Ipsum gt Or Ihr leuchtendes UYG Unfall Beinbruch 28 05 2006 gt Fixmedikation Vorbild und Ihre immerw hrende Figure 4 14 Elexis 1 3 User Interface German We think with some small but important changes the look and feel of Elexis can be optimized Figure 4 15 shows some of the changes we applied We provided consistent looking icons changed background colors and adjusted the way forms are displayed an example of this is the patient list on the left 46 Datei Bearbeiten Fenster Hilfe QS BRE ebe d t
31. Archie A Statistics Framework For Elexis 7 Technical Report Dennis Schenk Peter Siska March 2009 Supervised by Prof Dr Oscar Nierstrasz David R thlisberger University of Bern Switzerland Software Composition Group Abstract Archie is a statistics framework for the electronic medial records system Elexis Archie empowers Elexis to generically create anything from sim ple overviews to complex statistical reports about any data found within the Elexis system Depending on which plug ins are installed an Elexis instal lation contains data about patient demographics and history consultations drug administration practice management and inventory finances and ac counting laboratory etc Archie provides a platform for Elexis and for all installed plug ins to easily and rapidly create statistical reports without hav ing to be concerned with recurring aspects such as data input and output form validation result presentation or the user interface in general Data visualization is handled entirely by Archie it just requires the raw data to adhere to a defined standard Contents Introduction 1 1 Problem Statement o o o 1 2 Goalsand Challenges Soes sau o s ara 1 3 EMR Systems and Elexis cc LA ANTENE a o ane Wa Ki A Ee Ma bed eh wed oR 15 Structure of this Document s sess kua vaaka eu ri Domain E o lt dan i e eet en E REESE BH 22 TES QUO eg re Te E A E e A 23 Bas A ke e au dw ee A
32. Design and Evolution 3 1 Building A Prototype s a Ze ev hs es oe eH E d 3 2 Ideas for Architecture lt e s Se bu aa bk ea EO 3 3 Architectural and Design Decisions 3A Evolution E 3 9 Visualiz tion and EXPO seu diy bed bins mod oot aa Implementation GH ALOCE hera NEE Ra Rw a EOR 478 4 1 1 Structure of an Eclipse Plug in 11 2 Archie Plugin Definition s 2 lt sd os G4 OS a 4 1 3 Extension Point Definition 4 1 4 Initial Extension Schema 200 413 Improving Usability 2 22 4 rra 84 4 1 6 Extension Point Implementation 4 2 Data Providers and Datasets 222 nn 4 2 1 Content and Label Providers 4 2 2 Parameterization through Annotations 4 2 3 Provider Implementations o ANS o oe BH eS A A AA mi 4 3 1 New Statistics Action x lt vu a as wa 4 5 aa 132 Additional Gerbe serc acs he soa ee ee OT A 10 12 13 13 14 16 18 20 da Managing Classes lt lt k ki ice ala ds we e ee Be a 4 5 Factories oo m onen AO MATES e e doy e ee ge ee ee nE eo 461 Pie Charts s be gme Kaas Beare dhs we ae ee 4 6 2 Bar and Line Charts 4 6 3 The Chart Model AT Helper Classes o ered wa and mn Bed osta Ra aan 4 8 User Interface 2 2 2 0 u a a a k ka es 4 9 Elexis User Interface Contribution se 4 10 Dashboard Charts 22 44 resorts a suon Ead AV Tute l
33. Evolution We want to create a piece of software that will have an active userbase after releasing it Based on the points mentioned in the previous chapter EMR systems will most likely become more and more popular which is why we decided to create an application that is part of such an EMR system The following sections describe our initial idea for this project and its evolution as well as some important design desicions 3 1 Building A Prototype Before we started working on Archie in order to improve our knowledge of the fairly extensive Eclipse RCP framework we decided to create a simple prototype of a basic Eclipse application Sanclipse The functionality of Sanclipse as seen in Figure 3 1 is very limited Its purpose is to simulate a rudimentary EMR system using the Eclipse RCP framework The main window is divided into two parts the left hand side containing a list of fictive users or patients in the system the right hand side displaying details of selected users such as both names and the gender A user of the system can add new patients or remove existing ones from the list by using actions residing in the toolbar on the top Although the functions of Sanclipse are rather scarce we learned about im portant methods and core concepts of the Eclipse API Not only through the entire prototyping process but also by developing Sanclipse according to the tutorial of creating a chat application called Hyperbola as described thro
34. ages of EMR With paper based systems patient information and health records are dis tributed over many locations and are usually not readily available This leads to multiple records about the same aspect of patient health being kept at different facilities like hospitals doctors practices etc Redundant med ical tests have to be made for which results are probably already stored somewhere else Another factor is that patients most of the times can t pro vide full medical history about themselves Most people don t remember or don t know all their medical details If they think they do they maybe re member them wrong All this can lead to loss of time and money through inefficiency unnecessary administrative costs false diagnosis and errors in treatment Patient records should be available quickly and from all locations particu larly for chronic patients or in case of emergencies This leads to less spending of valuable time on tests and clarifications about a patients condition and medical history Systems which meet such requirements lead to better treatments less errors instantly transferable records complete information better reactivity and less administrative costs The numbers of doctors with own practices using EMR vary between 8 and 10 in Switzerland In the mentioned e Health Strategy of Switzerland it is discussed that by the year 2010 about 50 of doctors with own practice should deploy EMR systems and in 2
35. alid Filename Proposal During CSV Export 42 Widgets in the UI kin wu ee Harrer 44 Class hierarchy diagramm of UI Widgets L 45 Elexis 1 3 User Interface German 46 Elexis 1 4 User Interface German 47 Dashboard Charts Class Diagram 48 Possible solution for showing totals in the results table 58 Dashboard View lt gt sa ka e a ea aaa 62 The Sidebar Viewed ee ee ae A e 63 The Output View NN i OR a E ae de OH HP 64 Chart Wizard Choose Chart Type 65 Chart Wizard Bar Chart Page e e as een e e aa a Er 66 Chart Wizard Choose Rows cc 66 Example chart created with Chart Wizard 67 75 Listings 4 1 Extension Point Definition in plugin xml 22 4 2 User Overview Extension Point Implementation 25 4 3 Abstract Data Provider Methods 27 4 4 GetProperty Annotation Definition 30 4 5 Chart Model Object Variables 2 222220020 41 46 Example Annotation osorno Kw 50 C 1 Methods to override 2 2 Cr nn nennen 69 Gu createHleadings Example rajar ae ara en 70 C 3 reateContent Example lt lt s sa sa db erden 70 C 4 Annotated method pair for additional statistic parameters 72 Ch Registering a statistic with the Archie Extension Point 73 76 Bibliography Arth08 Bell04 Bhen06 Coch89 Gall02 Gesu07 Laff06 McA f07 Vale04
36. and table factories work and How is the UI to model interaction handled The following section described how the initial design evolved during the implementation process 3 4 Evolution At first we create a DataSet class which holds any data in a matrix like form and simplifies conversion to textual table and chart based output As we progress the fact that the dataset can contain any data leads to problems in the UI when trying to sort results tables Because of this we continue refactoring Datasets by adding another constraint Datasets should only be able to contain objects that implement the Comparable interface This means 18 that for some classes that do not implement said interface developers have to create a wrapper class We did this for example for the Elexis class Person with the PersonWrapper class in the samples project to enable alphabetical ordering See Section 4 2 for more details on the final implementation of DataSet One of the bigger refinements is the introduction of actions as user input and controlling mechanism Read more about our final implementation in Sec tion 4 3 Following this introduction the concepts of Table and ChartFactory lost their initial meaning of controlling entities The concepts of these two classes are converted to less dominant roles namely as data creation factories described in Section 4 5 Most refactoring efforts and changes happen in the UI part though The reason for thi
37. ap them in the appropriate model classes in order for Archie to use it A more detailed description on how to implement your own data providers can be found in Section C 1 4 3 Controllers Most of the controller part of Archie is implemented by using actions The term action in Eclipse refers to a visible element that allows users to initiate a unit of work McAf07 Given this definition actions in Eclipse are tightly integrated into the UI Eclipse defines several different extension points for several kinds of actions meaning that actions can be contributed to the UI either by definition using extension points but also programmatically using code In our case the actions always belongs to a view a part of the user interface thus are being defined programmatically Archie defines a couple of actions some of them for starting a new statistical query opening the chart wizard starting the chart creation in the dashboard or exporting results of a data provider An overview of these classes can be seen in Figure 4 6 We describe them further in the following sections http java sun com javase 6 docs api java util regex Pattern html 32 CreateChartsAction ChartWizardAction RefreshChartsAction dashboard Dashboard WIZARD_PAGE_HEIGHT int dashboard Dashboard CreateChartsAction WIZARD_PAGE_WIDTH int RefreshChartsAction Override run ChartWizardAction Override run
38. atient It offers clinical decision support controlled medical vocabulary computerized provider order entry pharmacy and clin ical documentation applications It supports the patients electronic medical record across inpatient and outpatient environments and is used by health care practitioners to document monitor and manage health care delivery within a care delivery organization CDO The data in the EMR is the legal record of what happened to the patient during his encounter at the CDO and is owned by the CDO A Schematic concept of a sample EMR can be seen in Figure 2 1 Transaction Billing Eligibility Transferals Registration Lab Results Diagnoses Treat t L Patient N a SES Clinical Management Figure 2 1 Schematic concept of a sample EMR An Electronic Health Record EHR is a longitudinal record of a pa tients health over time and can be seen as a subset of each care delivery organizations EMR It provides clinician and also consumer access to clinical details captured from one or more encounters from different facilities It mostly spans over a full lifetime from birth to death An EHR itself requires the presence of EMRs because the EMRs are the actual data providers for EHRs in the form of EMR summaries EHR data is owned by the patient EHR contains data from episodes of care across multiple CDOs within a community region or in some instances the entire country A Clinical Information S
39. be managed confidentially and secured from unauthorized access Because of crosslinking of digital media over networks such as the Internet multiple access points are present facilitating interception of patient data Skepticism about the security of electronic patient records is high on both the patients and the medical institutions side Another problem is preservation of patient data and the mandatory ability of patients to be able to have access to all their medical data at all time Data has to be kept for long periods of time without data degradation This too creates new challenges for electronic systems and their acceptance 11 2 3 Elexis Elexis was first conceived in 2006 by Gerry Weirich a general practitioner and internist Before he decided to start from scratch with an open source Eclipse based project he was involved in several other small EMR projects Elexis was first developed by Weirich alone and was deployed in his own practice Requirements and feature requests were created out of his own needs Soon other people started to get aware of Elexis and more physicians became interested in the project One aspect which sets Elexis apart from other EMR systems is that it is mainly developed by actual practitioners not by software engineers Also requirements bug fixing request etc come directly from facilities where it is in productional use This gives Elexis credibility and closeness with and to its users Elexis g
40. c type of object By providing accessor meth ods for the stored instance it can be accessed by different classes mostly controllers at different times while ensuring that the instance of the object 36 ChartModelManager ProviderManager TableManager INSTANCE ChartModelManager INSTANCE ProviderManager INSTANCE TableManager barChartModel ChartModel provider AbstractDataProvider table Table pieChartModel ChartModel ProviderManager TableManager ChartModelManager getinstance getInstance getInstance getProvider setTable getBarChartModel setProvider getSortDirection setBarChartModel hasProvider getSortColumn setChartModel hasTable getPieChartModel setPieChartModel hasChartModel hasPieChartModel hasBarChartModel clean Figure 4 8 UML overview of the manager classes remains the same We differentiate between the following three manager objects in Archie ProviderManager This class manages the currently active data provider instance When a user selects and runs a particular statistic from the list the data provider for that statistic is set in the ProviderManager It is reused in some controller classes such as in NewStatisticsAction in ProviderlnformationPanel a user interface class or in the chart generation wizard TableManager This is a manager class for the Eclips
41. cally A good practice is also to initialize the member variable with a value If you default it to true for example the checkbox is selected in the UI from the beginning otherwise it is not Now let us take a closer look at the annotations The name is an identifier for your method pairs it is important that they are the same in both methods It also gets used as label displayed in front of the control in the UI The index parameter controls in which order your properties will be displayed in the UI as well as processed when they are set If you have several properties they get sorted in descending order based on the index With widgetType you can specify what kind of widget will be displayed in the UI For available widget types please consult ch unibe iam scg archie ui WidgetTypes The most important are 72 TEXT A vanilla text widget which can validate itself TEXT_NUMERIC A text widget only for digits TEXT_DATE A date widget which has a inline datepicker for easy date setting BUTTON CHECKBOX A simple checkbox is either selected or not TEXT_NUMERIC makes sure that only numbers are entered TEXT_DATE does the same for dates If you want different more specific validations for text widgets you can also specify a regular expression for it With validationRegex you can specify a regular expression pattern If the pattern will match the contents of a text widget a green tick will be shown symbolising that the widget
42. cg archie model AbstractDataProvider Browse amp class 18 name S E RS Implements Browse Figure 4 1 Extension Schema Definition Screenshot Initially our extension schema defines that an extension point can contain one or more elements of the type DataProvider whereas each DataProvider element needs to have a name and a class specified Furthermore the class definition in an element provides a constraint that asserts classes used in those definitions are subclasses of AbstractDataProvider as illustrated in Figure 4 1 Although the extension point schema definitions are written in XML the use of the graphical editor is encouraged Laff06 that is why the file syntax won t be explained further 4 1 5 Improving Usability The initial extension schema implementation does not allow for grouping of data provider names The name of each provider is listed in the user interface in acombo box Our samples plug in consists of seven different plug ins each with a different name Towards the end of the project Gerry Weirich lead 23 developer of Elexis contributed his own set of data providers in a custom plug in for Archie The list of data providers as presented to the user got confusing and messy which eventually lead to a lower usability Although we had an autocompletion mechanism defined and in place to fa cilitate the searching of available providers in the list we decided to add the possibility to group certain data provid
43. ctly from http code google com p archie downloads list Unpack the contents of the archive file into the plug ins directory inside your Elexis installation directory Note The archive should contain two jar files they should end up in the plug ins directory along with the files already present Download the demo database from http www elexis ch files demoDB zip and extract it to your Elexis installation folder Note the current demo database only contains entries for one month namely August 2008 Statistics gueries which use date ranges outside of this month return no results Start Elexis with the nl de CH parameter On Mac OS X systems issue the following command in a terminal window lt path to your elexis folder gt elexis app Contents MacOS elexis nl de CH On other oper ating systems add the nl de CH parameter to the elexis executable when running it Login with user name test and password test 60 Note In order to be able to set access privileges for Archie in Elexis either the system locale has to be set to de CH or Elexis needs to be started with the nl de_CH parameter This is a known limitation of the Elexis 1 4 version as described in Section 4 11 This bug will be fixed in Elexis version 2 0 6 To use Archie you have to give the test user admin privileges To do so open the preferences select groups and rights and access control click on Archie access select the
44. e SWT table dis played as a result in the main view We use it to store the current table instance in order to be able to reuse its sorting attributes in the chart creation wizard ChartModelManager This class manages the last used chart model We implemented it in order to properly encapsulate and reuse the model and its attributes when a new chart wizard is started This way a user does not have to but of course can enter all chart parameters again when creating new charts especially when only a few parameters need to be changed 4 5 Factories There are two different situations where we have to create certain types of objects based on a specific object or its properties Particularly in Archie 37 we create either tables or charts In order to group these repetitive tasks we created two factory classes as can be seen in Figure 4 9 TableFactory ProviderChartFactory INSTANCE TableFactory INSTANCE ProviderChartFactory TableFactory model ChartModel getInstance ProviderChartFactory createTableFromData getinstance createTable setChartModel createColumns createChart addColumnSort createPieChart createJFreePieChart createJFreePieDataset createBarChart createJFreeBarDataset SuppressWarnings createJFreeBarChart Figure 4 9 UML overview of the factory classes TableFactory Creates an Eclipse JFace TableViewer containi
45. e ability to make use of the Eclipse Jobs API and control the state of every chart creation Most important for us was to provide the ability to start a chart creation on demand Running the chart creators contained in the four dashboard charts on slower computers takes a lot of time Additionally these creators block the entire UI when Archie launches Implementing the chart creators as Eclipse jobs allows us to start the chart generation on demand as well 48 as cancel already running chart creators This also solves the UI blocking problem 4 11 Limitations Although we were able to realize most of the features we or the Elexis users wanted to have there are a few limitations to Archie These limitations can be divided into two groups internal and external Internal limitations result from our implementation approach external limitations are caused either by the implementation of Elexis or the Eclipse API Some of these limitations are described in the following sections 4 11 1 Internal Limited Widget Types The list of available widget types is modelled as an enum Enums in Java are a special type instead of being a class that represent a fixed set of constants This means the number of widget types in Archie is limited to the types described in Section 4 8 Although a plug in for Archie theoretically could implement its own widgets by extending the AbstractWidget class enums cannot be subclassed so a plug in cannot add its own t
46. e at http code google com p archie If you want to get information on how to develop for Elexis in general please consult the developer section on the Elexis web site at http www elexis ch This manual assumes that you have basic knowledge in Java Eclipse and Elexis development C 1 Extending AbstractDataProvider The first step is to create a class in your project which extends ch unibe iam scg archie model AbstractDataProvider If your statistics needs to constrain itself to a timespan you can also directly extend ch unibe iam scg archie model Abstract TimeSeries which extends AbstractDataProvider and is already set up for the task of handling time spans It has instance variables startDate and endDate which 68 you can use How to use these two variables in your code is up to you Abstract TimeSeries just sets up the UI in the right way and provides the two dates the user entered In any case you will have to override the following methods Listing C 1 Methods to override Jos x Return an appropriate description public String getDescription jas x Create dataset headings in this method protected List lt String gt createHeadings x Compose the contents of a dataset here protected IStatus createContent IProgressMonitor monitor C 1 1 Constructor In the constructor you set the name of your statistics job and you can carry out any additional initialization you might need C 1 2 g
47. e commonly used Java Standard Widget Toolkit SWT elements such as labels and separators 4 8 User Interface The entire user interface is built on top of the Standard Widget Toolkit SWT used throughout the Eclipse API Although most of the components used are standard SWT containers and widgets in order to provide a nice user feedback upon parameter setting in the parameters panel we created custom widgets that handle the user input An example can be seen in Figure 4 12 The superclass for every such widget is the AbstractWidget class an abstract class extending an SWT composite which makes it usable in other SWT components throughout the UI Each AbstractWidget consists of a label and a control controls such as simple text fields combo boxes or similar are parts of SWT Concrete implementations of abstract widgets provide methods for setting and retrieving the value of a widget as well as in certain implementing classes validation and quick fix methods These widget types as well as the validation parameters can be set in the data providers implementing plug ins by using annotations and regex patterns Annotations are described in Section 4 2 2 regex patterns in Section C 2 43 Start Date 01 01 2009 v End Date 31 12 2009 Cohort Size a5 v g Active Mandator Only M Abfrage Starten Figure 4 12 Widgets in the UI This gives each plug in total control over how the corresponding UI widgets to
48. er triggers the action from the UI in our case by clicking the Start Query button in the sidebar The run method in NewStatisticsAction performs a series of actions related to the currently selected data provider First it takes care of checking the validity of provider parameters and sets them on the provider if the parameters are valid Then the action registers itself as a job listener on the data provider and updates and populates the main view with information about the currently selected provider and its parameters Finally the new statistics action tells the data provider to run In this action we use the Observer Pattern as well as the lJobChangeListener interface of the Eclipse API On the one hand the action observes the ProviderManager which propagates an update event every time the currently managed provider changes By default the new statistics action is disabled As soon as the provider manager has set a valid provider the action enables itself and users can start statistical queries Additionally if at any time during the runtime of the application the provider changes and or is not valid anymore the action disables itself Thanks to the lJobChangeListener interface the action can react on different states of a job which in our case is the data provider We namely use the done final JobChangeEvent event method which is called after a job has terminated successfully or unsuccessfully The action populates t
49. erage profits per consultation grouped by age groups cohorts Start Date 01 01 2008 v 5 End Date 31 12 2008 v 5 Cohort Size s v Active Mandator Only V gy Start Query Cancel Figure B 2 The Sidebar View In the Sidebar View you can find all the available statistics in your Elexis system in a drop down list If there are too many statistics to scroll through you can also type in the name of a statistic and an auto completion helps you find it If you chose a statistic additional options and settings are shown in the sidebar When you have made your choices click on Start Query and a background job will be started to assemble the data B 2 3 The Output View When a statistics creation job is finished the results will be shown in the Output View Here you can sort the results by clicking on any table heading If you want to export the data to a Comma Separated Values CSV file you can do so by clicking on the Export Action button in the upper right corner of the view After exporting the data you can then open it for further processing in Microsoft Excel Open Office Calc or any other software which supports CSV files If you want to draw the results as a graphical chart click on the Chart Wizard action button on the upper right corner of the Output View 63 x Dashboard Output view gt oo Current Statistics Consultations Statistics Start Date 01 01 2008
50. ers that belong to the same plug in or to the same family of tasks The simplest and in terms of usability yet effective solution is to prefix every data provider name in the list with a defined term There are two possibilities for the definition of that prefix One of which is by using a term based on the fairly cryptic ID of each plug in contributing to the data provider extension point such as ch unibe iam scg archie samples The first part of the plug in ID would be obsolete in the UI so we need to manipulate the IDs to only retrieve the latest portion of it in order to get a nice human readable prefix such as Samples The other variant imple mented in the end is to give developers the possibility to define an optional prefix in a handy way by using the same extension schema as for their data providers Verf gbare Statistiken Verf gbare Statistiken Verordnungs bersicht Samples Verordnungs bersicht v Offene Posten Buchhaltung Rechnungen nach F lligkeitsdatum Buchhaltung Zahlungsjournal Samples Benutzer bersicht Konsultationen Zeit Faktura Journal Patienten Kosten Umsatz und Gewinn Samples Diagnosen Verordnungs bersicht Samples Konsultationen Zahlungsjournal Start Date 01 01 2009 v End Date 31 12 2009 v E End Date 31 12 2009 v E Abfrage Starten 3 Abbrechen 4 Abfrage Starten Abbrechen Figure 4 2 List of Providers Before and After the Extensio
51. etDescription This method returns a short description of your statistic This description gets displayed in the Archie sidebar when your statistic is selected It should convey what your statistic is about in general and what kind of results it will return C 1 3 createHeadings This method returns a list with strings The list will be used to populate the table headings in the result view For example 69 Listing C 2 createHeadings Example JET x Osee ch unibe iam scg archie model AbstractDataProvider createHeadings Override protected List lt String gt createHeadings final ArrayList lt String gt headings new ArrayList lt String gt 2 headings add Heading 1 headings add Heading 2 return headings Pay attention to the number of headings you return it has to correspond to the number of columns you create in createContent CIA createContent This is the method that does the actual work of the statistic Let us look at an example Listing C 3 createContent Example inheritDoc Override public IStatus createContent IProgressMonitor monitor Initialize result list final List lt Comparable lt gt gt result new ArrayList lt Comparable lt gt gt Set job size and begin task int size allPatients size monitor beginTask Working on patients ldots size for Patient patient allPatients Check for Cancellation if monitor isCanceled
52. ethod annotated with a x SetProperty method with the same value to have any effect on the data provider This name will also be the title of the corresponding widget x shown in the UI public String name fa x A brief description about the getter method x This is used for toolips of the UI elements public String description default x Property index Defines the order in which this property is displayed and processed public int index default 1 Widget type What kind of widget x public Widget Types widgetType default Widget Types TEXT ja x A regular expression pattern string to be performed as validations on the given method public String validationRegex default 30 x A validation error message that will be displayed upon unsuccessful x validation of the input for the method containing this annotation public String validationMessage default In addition to the index and name variables the GetProperty annotation is composed by couple of options Each of these options is described in more detail in Listing 4 4 Not only can the annotations for getter methods contain a description used in the UI to provide a more detailed feedback about a provider parameter but getter annotations also take care of a proper validation expression and message as well as the widget type with which they should be presented to the user Start
53. ework should have Slowly but surely the requirements for our projects became clearer 54 After the initial exchange of ideas and requirement proposals we started to write Archie From time to time we presented our advancements and got feedback on them This was also a good way to see if we were still on track with the project The feedback we got was mostly very good and we were glad to find that the Elexis community is albeit a small one very active and constructive At the last stages of development we started working closer with Gerry Weirich which was very fruitful He started to use and test Elexis which gave us more important feedback Also we were able to ask questions about the Elexis code base directly to its creator and could thus improve and opti mize Archie to work better and closer with Elexis Subsequently the Archie framework was used by Gerry to create an account ing tool Weir09a It contains multiple analyses like an open bills statistic a payment journal statistic etc We were glad that our framework was already being used as a base for other projects and that we were told that it had been pretty easy to do so which was one of our goals with Archie Reception of the framework has been very good up until writing of this paper We feel that there is great interest from users and developers alike The plugin gets many feature requests and it will become part of the next major Elexis release 59 Chapter 7 Conclu
54. fen des Armes O Rechnungen F lle 23 Oso G Status proc Labor dann gt Bemerkungen Ineubeurt D KYG Krankheit Allgemein 22 11 gt Fixmedikation Figure 4 15 Elexis 1 4 User Interface German 4 10 Dashboard Charts In addition to the previously mentioned widgets for user input other user interface components with a custom implementation in Archie are the dash board charts Dashboard charts are customized SWT composites that consist of two im portant parts a ChartComposite being a JFreeChart class containing the actual chart that is displayed to the user and a AbstractDatasetCreator that is responsible for creating the dataset for each chart in our chart composite Again it is important to note that the dataset created by this creator is not the same dataset object as we use in our providers but that JFreeChart merely uses the same naming for their chart object s data The AbstractDatasetCreator object is very similar to our AbstractDataProvider Every dataset creator is a job so that the creation of each dataset can be moni tored outside of the creator more specifically by the AbstractChartComposite class being the job listener As soon as the dataset creator s job is done it 47 Eclipse API Classes N gt creator AbstractDatasetCreator chartComposite ChartComposite initializeCreator AbstractDatasetCreator initializeChart JFreeChart startCreator void cancelCreator vo
55. has been filled in correctly Else a warning or error symbol will be shown Here is an example validationRegex 1 9 1 d 0 2 If you provide a custom regex pattern you also have to provide a description of it using validationMessage This message will be displayed if the widget does not validate Here is an example corresponding to the one above validationMessage This field has to consist of at least one at most three numbers C 3 Registering with the Archie Extension Point After you have written your statistic you have to register it with the Archie Extension Point To do this you have to edit the plugin xml or fragment xml file of the Elexis plug in your statistic class resides in Just add and edit the following lines Listing C 5 Registering a statistic with the Archie Extension Point lt extension point ch unibe iam scg archie dataprovider gt lt DataProvider category name of your statistics category class name of your statistics class name Name of your Statistic gt lt DataProvider gt lt category id name of your statistics category 73 name some Category gt lt category gt lt extension gt The category is used to group similar statistics together For example all statistics dealing with some monetary aspect could be grouped in one cate gory After you recompile and restart Elexis your newly created statistic should now be in the drop
56. he element at the given columnIndez Label providers can also provide images for their elements Our default label provider implementation does not deliver any images for dataset elements this is why the method getColumnlmage returns null Subclasses of AbstractDataProvider can easily override these default provider implementations by setting their own content and label providers and thus controlling how the data they return will look like when presented to the user 4 2 2 Parameterization through Annotations An important question that came up during the refinement of our initial de sign was how parameters for data providers whose implementation we not necessarily know as Archie can and should be extended by anyone could 29 be set in our framework We solved this particular problem by annotating getter and setter methods for data provider parameters with custom anno tations Particularly Archie provides two kinds of method annotations to use in data providers GetProperty for annotating getter methods and SetProperty for annotating setters Each get and set property pair is associated by a name and they both have an index variable that can be set Index controls the order in which the getters and setters are called by the controller or UI Listing 4 4 GetProperty Annotation Definition Retention RetentionPolicy RUNTIME OTarget Element Type METHOD public interface GetProperty Property name There has to be a setter m
57. he result view with results from the data provider This can either be a table contain ing the data from a dataset of a provider or an appropriate message for the user if the query did not return any results The sequence diagram shown in Figure 4 7 demonstrates how the action interacts with the data provider as well as important controller objects and result panel Bell04 34 erManager results ResultPanel ChartModelManager action NewStatisticsAction 1 setEnabled false 2 getinstance provider AbstractDataProvider 3 getProvider Main run method used in N S actions to start them 4 schedule 5 addjobChangeListener action Invoked when the data N provider job is done as the action is registered as listener 6 getinstance 7 getProvider 8 getinstance 10 showEmptyMessage 12 setEnablddtrue More classes are involved here e g N TableManager for creating the result table but basically if the dataset has data the results are displayed in the result panel Figure 4 7 Data Provider Controller Interaction 35 4 3 2 Additional Actions As previously mentioned in Section 4 3 there are a few additional actions we use in Archie other than the new statistics action The following list shows and describes these actions used as controllers in Archie ChartWizardAction This action is defined in the main view where the res
58. id TT 1 En initializeChart JFreeChart initializeCreator AbgtractDatasetCre bor lt lt lInterface gt gt lJobChangelListener done event JobChangeEvent void AbstractDatasetCreator dataset AbstractDataset createContent monitor IProgressMonitor IStatus run monitor IProgressMonitor IStatus ConsultationMonkyDatasetCreator createContent monitor ProgressMonitor Status initializeChart JFreeChart initializeCreator AbstractDatasetCre PatientsDatasetCreator createContent monitor PrograssMoniar States AgeHistogrammChart initializeChart JFreeChart initializeCreator AbstractDatasetCreator AgeHistogrammDatasetCreator createContent monitor IProgressMonitor IStatus Figure 4 16 Dashboard Charts Class Diagram removes the initially displayed working message and shows the created chart based on the creator s dataset Every implementing class of a chart composite needs to provide its own par ticular dataset creator based on the type of the chart the composite should display A detailed illustration about the class dashboard charts class hier archy is depicted in Figure 4 16 The big advantage of this approach is not only the ability to monitor the creation of the charts along with their datasets which depending on the size of the dataset can be time consuming but also th
59. ime However on the other hand it is 56 this very amount of blocks that makes it difficult to use The documentation on some parts of Eclipse is very scarce often your best source is a good web search engine or the Eclipse newsgroups Another lesson was that working in a team often requires adapting the team organization to the current task As an example for us when writing code it was most efficient to do it in pair as described in Chapter 5 albeit when writing the thesis it was best to work individually on separate chapters or sections 7 2 Future Work In this section we outline future work and possible additional functionalities of Archie Some of the following paragraphs are related to the limitations described in Section 4 11 Flexible Widget Types Archie currently supports only a fixed set of UI widgets used for user input The widget types could be implemented as a class instead of an enum With this an implementing plug in can extend the available widgets by adding its own widget types such as radio drop down or other custom widgets Another approach would be to define a new extension point in the Archie plug in definition file and let implementing classes use this hook to provide custom widget types Totals in Result Table For several statistics users might be interested in a computed total of each column This could be implemented by extending the dataset model and adding a special row Each cell of this additional row co
60. ma Plug ins can publish these extension points and thus specify what an implementation needs to fulfill in order to be usable by the plug in Archie only defines one extension point The definition contains an unique identifier ch unibe iam scg archie dataprovider a human readable name and a reference to a extension point definition schema as shown in Listing 4 1 As 22 indicated by its name this extension point allows other plug ins to contribute their statistical data to Archie 4 1 4 Initial Extension Schema The other part needed for an extension point in Eclipse to work is an ex tension schema as referenced by the extension point definition in plugin xml from the previous Section 4 1 3 Basically an extension schema contains an XML formed definition of elements and attributes that extensions to that extension point implementors must declare Furthermore the extension schema can contain implementation examples API descriptions and addi tional documentation or licensing information Extension Point Elements E Attribute Details Specify the XML elements and attributes which are allowed in this extension Properties for the class attribute Name class v del extension New Element W Sequence 1 A Deprecated O true false d i DataProvider i a N New Attribute Use required 8 point a id New Choice Type java 4 name H v dEl DataProvider 2 wen Py Extends nibe iam s
61. n Point Changes We extend the initial extension schema by adding a new optiona element called category A category consists two attributes a unigue identifier id and a name Every ch unibe iam scg archie dataprovider extension point im plementation can now define one or more categories and assign them to the DataProvider elements in the definition The name of a category is then ren dered as a prefix to each data provider in the list As an example shown 24 in Figure 4 2 all our data providers in the samples plug in are in the Sam ples category so that they show up as Samples Some Provider Name in the list Grouping of other core Eclipse plug ins such as views or perspectives is defined equally Another advantage of this approach is that defined categories are available across plug ins allowing developers to create provider families and group together providers across different plug ins only if plug ins that need to share categories are both included in Eclipse run configurations or application distributions 4 1 6 Extension Point Implementation Any plug in can now implement the defined extension point We do this in our Eclipse fragment ch unibe iam scg archie samples for every class that provides statistical data to the main plug in Listing 4 2 User Overview Extension Point Implementation lt xml version 1 0 encoding UTF 8 gt lt eclipse version 3 2 gt lt fragment gt lt extension point
62. n The status of the current job x Osee org eclipse core runtime IStatus protected abstract IStatus createContent IProgressMonitor monitor In addition to access methods for providers and dataset the AbstractDataProvider class contracts its subclasses to implement the abstractly defined methods shown in Listing 4 3 Because a data provider extends an Eclipse job it can be scheduled for execution Upon the start of a provider its inherited run IProgressMonitor monitor method then populates the datasets headings with the result from the createHeadings method and triggers the createContent IProgressMonitor monitor method where subclasses need to initialize the datasets content and return a status of the execution signalizing its outcome The monitor passed to the creation method allows for propagation of the progress of a running job to the UI However it s the responsibility of each subclass to implement the monitoring of a data provider properly An UML overview of the discussed classes can be seen in Figure 4 4 21 QueryContentProvider dataSet DataSet QueryContentProvider getElements use dispose inputChanged use ColumnSorterAdapter viewer TableViewer index int ColumnSorterAdapter Override widgetSelected use ColumnSorter sortDirection int index int ColumnSorter SuppressWarnings Override compare setSortDirecti
63. name Proposal During CSV Export One example application of such a helper class can be found in the CSV export functionality In the popup dialog presented to the user where he can type in the filename used to save the data a filename proposal is al ready present An illustration of this is shown in Figure 4 11 This filename is a cleaned up string composed by the name of the data provider and the current date containing only valid filename characters Moreover all whites pace characters are replaced by underscores This string is generated by the 42 StringHelper class In addition to this StringHelper class the following list shows all other helper classes available in Archie Array Utils Provides convenience methods for printing out as well as check ing the existence of certain elements in arrays DatabaseHelper This class is used in the dashboard overview composite in the UI and provides methods for easy access to certain data in the database DatasetHelper Provides methods to sort datasets to type check dataset columns and to retrieve column indexes based on the column title ProviderHelper This class offers methods to retrieve all annotated meth ods and thus parameters for a data provider as well as to retrieve and set values for these methods Furthermore it takes care of sorting and storing these methods in a map for further processing SWTUtils Offers convenience methods for user interface classes to easily creat
64. ng a Table ob ject based on the data from a given dataset The factory method create TableFromData Composite parent DataSet dataset IL abelProvider labelProvider IContentProvider contentProvider also takes two providers label and content as parameters as well as the Composite object the table gets displayed in Context menus and other enhancements can only be added to viewer contain tables rather than table objects which is why we decided to return the TableViewer rather than the Table object during the devel opment of the project as opposed to our initial implementation ProviderChartFactory This class creates ChartComposites based on a given ChartModel object as described in Section 4 6 3 This also means taking care of mapping the definition of our dataset into a particular definition of a JFreeChart dataset as described in Section 4 6 Both factories are implemented as Singleton although they could also be implemented as classes containing only static methods thus not have an in stance 38 4 6 Charts We want to give users a means to evaluate the result of a data provider not only by consulting the result table in the UI but also by providing a convenient means to generate graphical charts In order to achieve this in a general way one that is in line with displaying data not known upfront we created a chart model and a chart wizard The data displayed in the results table can be exported into a graphical chart by usi
65. ng the chart wizard in the result view There are three chart types that can be created using this wizard pie bar and line charts All of our charts are rendered using the free and feature rich Java chart library JFreeChart described in Section 3 5 JFreeChart provides a well documented API and supports a wide range of chart types out of which we only use the three types mentioned before A chart object in JFreeChart is based on a dataset this is a different dataset object than ours but the term dataset in JFreeChart refers to something very similar a collection of data keys and values This chart object can be used to create a ChartComposite which is a subclass of an ordinary SWT Composite and thus can be used and displayed in other composites and containers in the UI For each chart type we need to create a factory method which maps the values in our dataset into an particular dataset definition in JFreeChart in order to be used to create a chart 4 6 1 Pie Charts Pie charts in JFreeChart are based on datasets that implement the PieDataset interface JFreeChart provides a default implementation of this interface in a class called DefaultPieDataset This dataset is a mapping of keys to values As each row in the JFreeChart DefaultPieDataset is a row in our dataset definition we only had to create the mapping between these two datasets by defining the keys and values which in our case corresponds to picking one column from ou
66. ntains the totalised value of all corresponding cells in a numeric column column that contains numeric values only as totalising make most sense for numeric values Such a row needs to be excluded from the sorting of a dataset which is why it is marked as special in the dataset model Figure 7 1 shows how a possible solution for showing totals in the results table could look like User Interaction in Result Table The misssing user interaction could be implemented by allowing plug ins to provide custom selection listeners Currently every plug in data provider can define a custom content and 57 Original Codesystem Amount Kosten Umsatz Gewinne Eigenartikel 7 CHF 0 00 CHF 53 04 CHF 53 04 Laborleistung 2188 CHF 0 00 CHF 25 846 20 CHF 25 846 20 Medicals 49 CHF 1 000 56 CHF 1 299 62 CHF 299 06 Medikamente 1014 CHF 11 343 24 CHF 19 751 64 CHF 8 408 40 Tarmed 6449 CHF 0 00 CHF 93 799 24 CHF 93 799 24 Proposed Codesystem Amount Kosten Umsatz Gewinne Eigenartikel 7 CHF 0 00 CHF 53 04 CHF 53 04 Laborleistung 2188 CHF 0 00 CHF 25 846 20 CHF 25 846 20 Medicals 49 CHF 1 000 56 CHF 1 299 62 CHF 299 06 Medikamente 1014 CHF 11 343 24 CHF 19 751 64 CHF 8 408 40 Tarmed 6449 CHF 0 00 CHF 93 799 24 CHF 93 799 24 9709 CHF 12 343 80 CHF 140 749 74 CHF 128 405 94 Figure 7 1 Possible solution for showing totals in the results table label provider These providers are respected when the results table is ren dered in the UI Using the same system a data
67. oblem domain clarify terms and definitions and report about the status quo of EMR In Chapter 3 we describe how we were going on about the tasks at hand how we designed the architecture of our plug in what technical challenges we faced and how we resolved them In Chapter 4 we present our final implementation and explain the main technical solutions as well as the architecture Chapter 5 shortly describes how we were organized as a team and how we split up work In Chapter 6 we present how we communicated with the users and developers of Elexis and how we incorporated their feedback into our work Chapter 7 provides information on what we learned from the project and delivers a short outlook on future activities In Appendix A we explain under which license Archie will be released and why In Appendix B and Appendix C we provide manuals for Archie users and for developers respectively on how to write new statistics on top of the framework Chapter 2 Domain In this chapter we take a closer look at the domain Elexis and Archie belong to the domain of electronical medical records Besides the technological aspects there are numerous other aspects mostly political that we find are important and worth a short analysis As mentioned in the introduction most physicians in Switzerland still use paper based patient records EMR adoption is slow but gaining ground In the following we study the disadvantages of paper based systems and the advant
68. ode but besides getting the data of interest out of the system the underlying framework takes care of the rest handling user interaction and preparation of data and its presentation Second we want to give users the possibility to quickly generate and export textual and graphical reports about any data in Elexis An example of a user requirement is to answer the question How much money have I earned over the last six months and what were my expenses over that time period The generated reports will provide answers to such questions Third we want to create a framework and accompanying statistics that are widely used by both Elexis developers and its users To accomplish this we work as close as possible with both sides We gather requirements feature requests ideas and also critique from the people who are using Elexis in practices Besides these three goals the framework has to be intuitive easy to use and understand and come with a nice looking user interface Challenges One challenge to achieve these proposed goals is the structure of the Elexis application It has grown into a large system with a lot of interdependent classes and methods The source code is often hard to read and understand because of bad formatting and naming mixing of German and English vocabulary just a few JavaDoc and other comments etc For these reasons it is difficult to find ones way through the program code and to find the right starting points
69. on getIndex setIndex isComparable dataSet DataSet AbstractDataProvider labelProvider ILabelProvider contentProvider IStructuredContentProvider AbstractDataProvider getDescription createHeadings createContent Override run execute getContentProvider getLabelProvider getDataSet initializeProviders set abelProvider setContentProvider SetDataException serialVersionUID long SetDataException use use DataSet content List lt Comparable gt headings List lt String gt width int DataSet DataSet AbstractTimeSeries startDate Calendar endDate Calendar AbstractTimeSeries initDates setStartDate setCell getStartDate getCell setEndDate getRow getEndDate setRow GetProperty metaGetStartDate getColumn SetProperty metaSetStartDate addRow GetProperty metaGetEndDate iterator SetProperty metaSetEndDate getContent setContent getHeadings setHeadings Override toString isEmpty Override clone Figure 4 4 AbstractDataProvider class and surroundings 28 4 2 1 Content and Label Providers Our initial definition of the entire application design already contained label and content pr
70. ot noticed in medical journalism and was sponsored to be presented at some conferences Medical equipment manufacturers started sponsoring development of equipment specific data importers Today Elexis is deployed in around 50 practices Weir09b It offers medical record functionality inventory management billing functionality and debtor control There are many plug ins available which offer functionalities ranging from laboratory equipment data import to several other import and export possibilities from financial extensions to general EMR extensions It provides interfaces to numerous medical portals and supports standards like HL7 SGAM eXchange and XML Invoice 4 0 From January 2008 to January 2009 Elexis was at the center of argo LEAD a project which was in its core a field test with numerous involved physicians to answer questions like how high are the costs of a switch from paper based patient management to an EMR where are the main difficulties how are the changes managed by facility employees and how practical is Elexis as a solution The project was a success and was evaluated As a result Argomed an organisation of medical practitioners with around 1000 members decided to contribute to a company that wants to assist with the distribution of Elexis and provide support to the current user base 3http www weirich ch http www rgw ch elexis dox argoLEAD pdf Shttp www argomed ch 12 Chapter 3 Design and
71. oviders as described in Section 3 3 what merely has changed during the development process are the names of those providers namely from DataContentProvider to QueryContentProvider for content providers and from DataLabelProvider to QueryLabelProvider for label providers These providers are the connection between the data in the model and its representation in the user interface as they control how the data from a data provider s DataSet is being rendered in the table in the result view Content providers need to implement the IContentProvider interface from the Eclipse API whereas label providers need to implement one of the appropri ate interfaces from the Eclipse API such as ILabelProvider ITableLabelProvider or CellLabelProvider In our implementation the default content provider in our data providers is a IStructuredContentProvider whereas the label provider is a simple implementation of an ITablel abelProvider Our content provider merely returns the stored dataset s content as an ar ray in the interface method getElements Object inputElement ignoring the inputElement passed to this function The label provider uses the interface method getColumnText Object element int columnIndex to return the corre sponding cell in the dataset In contrast to the content provider which ignores the inputElement parame ter the label provider takes the element parameter as a row in the dataset and returns the string representation of t
72. practices They showed us what system they had in place to manage their patients what their workflow was and what the advantages and disadvantages of these were We asked what they would expect from an EMR system and how they would go on about migrating This way we got an insight into the difficulties of migrating paper based systems but we also saw that human factors are ac tually more important Radically changing workflow for example can be very hard especially for doctors and employees of long existing practices With this and other input that was given we got a feeling for what was important and what challenges but also chances lie in the domain of EMR Later the idea of Archie was already born we visited a practice where Elexis was used We talked to the doctors using the system and got more informa tion on it in general and insight into how it changed workflow and practice management We also met Gerry Weirich the main developer of Elexis and talked with him about our ideas The next step was to contact more developers of Elexis and its user base which interestingly are mostly the same people the main developers all use Elexis in their own practices We registered ourselves on the Elexis mailing list and on the forum where we proposed our idea for Archie We got various feedback e g for what kind of statistics they would wish the most for From these ideas and wishes we started to draw out what capabilities our statistic fram
73. r dataset for each a keys column and a values column 39 4 6 2 Bar and Line Charts The implementation of bar and line charts in JFreeChart is slightly different to that of pie charts as the underlying dataset needs to implement a different interface namely the CategoryDataset interface A default implementation of this interface is provided in JFreeChart by the class DefaultCategoryDataset The difference between a bar and a pie chart lies in the possibility to plot more than one category column in a bar chart Consultations vs Costs rowTitleColumnIndex Value onsultationen Totale Kosten Altersgruppe U 85 CHF 642 03 30 35 87 CHF 552 78 35 40 81 CHF 536 70 40 45 100 CHF 646 42 m Totale Kosten m Konsultationen Figure 4 10 Dataset Mapping in Bar Charts We created the mapping by defining a variable rowTitleColumnlndex that holds the index of the column in our dataset which provides the title for each row and each category column in the category dataset as well as similarly to the pie chart mapping an index of all rows and columns categories that should be plotted The row titles based on the rowTitleColumnlndex variable are the labels on the x axis in a bar chart Line charts in JFreeChart are based on the same dataset as bar charts which is why the same definitions and mappings apply for line charts An example of the dataset mapping in bar charts is shown in Figure 4 10 4 6
74. ragment Host in their manifest file thus belongs to a plug in without which it cannot exist Moreover the extension point definitions are stored in a fragment xml file instead of plugin xml Arth08 Archie defines the following plug in structure in Eclipse ch unibe iam scg archie This is the main plug in containing the entire object model views and controllers definitions and implementations of Eclipse extension points and all read only data such as images and documentation ch unibe iam scg archie samples The samples plug in fragment contains sample statistics implementations to show how the main plug in and its framework can be used 4 1 3 Extension Point Definition The core of the entire Archie framework structure is the extension point defined in the main plug in This extension point allows other 3rd party plug ins or plug in fragments to interact with Archie Without other plug ins implementing this extension point Archie only has a very limited use Listing 4 1 Extension Point Definition in plugin xml lt xml version 1 0 encoding UTF 8 gt lt eclipse version 3 2 gt lt plugin gt lt extension point id ch unibe iam scg archie dataprovider name Archie Data Provider schema schema ch unibe iam scg archie dataprovider exsd gt lt plugin gt An extension point in Eclipse is a definition of a plug in interface and de scribes the configuration point using an XML sche
75. rt of each Eclipse plug in but unfortu nately not every part There are limitations of the Eclipse international ization mechanism namely when it comes to translating VM retained an notations that are being read at run time In Archie this is the case for parameters of a data provider Listing 4 6 Example Annotation Oparam currentMandatorOnly SetProperty name Active Mandator Only public void setCurrentMandatorOnly final boolean currentMandatorOnly this currentMandatorOnly currentMandatorOnly Each parameter of these annotations such as for example name in Listing 4 6 must have a value Translations in Eclipse work by defining constants in a Messages java file and associating these constants with their language specific values in a separate text file such as messages properties or in a dedicated language file such as messages_de properties for German The values of these constants of translatable strings are associated by Eclipse at runtime depending on the language set in the running application Anno tation parameter values need either to be concrete values or constants with defined values at compile time which is why we cannot use translations in annotation parameters Access Control Lists Archie uses the access control list ACL extension point definition ch elexis ACL Contribution provided by Elexis for restricting access to certain users to the entire Archie application This is particularly
76. s is probably that we didn t have a fleshed out concept for the UI and made it up as the project progressed One of the most important aspects is the introduction of annotated DataProvider methods Using Java annotations we are able to programm the UI generically as described in Section 4 2 2 This means that we have an UI that sets itself up with the right input fields just by looking at how the corresponding methods are annotated Another important aspect in the UI is the creation of generic input field classes for specific input types which are able to validate themselves and provide general user assistance as described in Section 4 8 Another aspect of the UI we had not thought about in detail was how the charts were being created from statistic results First we wanted to create chart generation somehow generically However we noticed very quickly that this went beyond the scope of our project and decided to create a chart creation wizard With such a wizard a user can choose what parts of the results he wants to use to generate a chart Moreover the chart wizard determines the type and look of the generated chart This is explained in more detail in Section 3 5 with implementation details in Section 4 6 Another thing that changed from our initial design was that AbstractDataProvider inherits directly from the abstract class Job of the Eclipse API instead of the Elexis BackgroundJob class In Eclipse 3 3 new Job API classes were intro duc
77. sions In Chapter 1 we stated the goals of this project Then we showed how to the data provider can be used for retrieving data about Elexis and how the data provider stores this data in a dataset modelled as a table for further processing We have shown how other Eclipse plug ins can hook into the Archie framework in order to provide their own statistics Moreover Archie provides a chart wizard that allows users to visualize statistical textual data by using pie bar or line charts as well as export this data in a standardized format for further processing Archie is written as an Eclipse plugin which allows the users to easily deploy as well as maintain it Furthermore we worked closely together with an active Elexis userbase as well as its main developer to ensure that the final product will be used on a day to day base by Elexis users 7 1 Lessons Learned We learned that contributing to an already existing project can often be difficult and time consuming Not only do you have to familiarize yourself with the available program code but with the programming style of the project owners as well The latter can have a great impact on how fast this familiarization process progresses Moreover we learned that Eclipse is a great Java framework It provides a very extensive set of API classes and methods that allow you to develop complex applications On one hand this great amount of already available building blocks can save you a lot of t
78. ssible statistical functions Elexis does not include the possibility for users to generate system wide and customizable statistical reports about patients drugs consultations or any similar aspect of recorded data Im portant analyses about state and development of the practice using Elexis are missing There is for example no easy way for doctors to see how many consultations they have given over a certain time frame or how this correlates with the practice s income and spending What Elexis misses is the ability to inform its users about the state a practice is in and also to visualize this data It should help in making predictions about future developments and to tell what services were provided Elexis needs the ability to inspect flow of money and inventory provide key facts about patients in the system Also it needs to be able to compare different aspects of such data This lack of basic overview and data correlation motivates the idea to de velop a statistics framework for Elexis a platform that helps to gather compose and also visualize overviews of any data in the system rapidly and easily Another aspect requiring improvement is usability and screen design We also contributed in this area to the project see Section 4 9 1 2 Goals and Challenges Goals First of all we want to create a framework for developers allow ing them to quickly and easily assemble their own statistics Programmers still have to write Java c
79. t ss ma dane sudo SIR wd bee oo fi RS AT Internal lt sas sia sm ae ee ee ee 41102 External ur a E re HR Oe ne sh Team Organization Requirements Engineering and Validation Conclusions 7 1 Lessons Learned 2 2 2 2 2 2 0 2 006 8c ee eee na 1 2 Future Work se 22 use Dune Ei Licensing of Archie User Manual BA Quek St os pois ee eee ee N lt a ee A B 2 1 Dashboard View B 2 2 The Sidebar View cc B23 The Output Vier 224 2 bee e ed KE a Developer Manual C 1 Extending AbstractDataProvider DAN CST oo a s aie mt e cat Ba eee eee ee KSE PD GNPiOA 22 4 244 Se OES eee EE MORES CLA Gesten lt lt lt e sos noei KEN ss e ee CLA IRENE ends A C 2 Adding Additional Parameters 4 C 3 Registering with the Archie Extension Point CA Where to Get Additional Help lt lt 6 sessa 4 A List of Figures Bibliography 54 56 56 57 59 60 60 61 61 63 63 68 68 69 69 69 70 71 73 74 75 77 Chapter 1 Introduction 1 1 Problem Statement Elexis is an open source Electronic Medical Records system detailed infor mation about Elexis and its history is provided in Chapter 2 It has grown into a complex software solution with about 850 classes around 100k lines of code and many features There are aspects that are yet not very mature though Despite a large and broad functionality and some crude and not very acce
80. t to use in your chart For bar charts for example you might want to sort the rows and only take the top ten to plot In our example we choose all rows as there are only five anyway See Figure B 6 When we click on Finished our chart will be created as the example chart in Figure B 7 shows You can further investigate the chart by zooming in or out on certain parts When you move your mouse around tool tips get displayed explaining what you see You are also able to change certain aspects of the chart such as colors captions etc by right clicking on it You can also save the image to your hard drive by clicking on Save As 64 Chart Wizard the Chart Wizard you can create a graphical representation of your statistic results on Next gt SIE Figure B 4 Chart Wizard Choose Chart Type 65 m lo x Choose Bar Chart Parameters Please specify parameters for your bar chart i Refine the type of your chart Line combine values bar charts separate them by category Bar Chart y Please choose which columns you want to include in the bar chart Each column depits a category the rows will be drawn per category You will be able to choose the rows in the next steps Name of the Chart Income Costs per Mc Select a column that will be used For the row labels bate i Chose which columns should be used for categories in this bar chart Row values will be grouped per category J
81. ted which maked the Elexis wrapper class BackgroundJob obsolete 19 3 5 Visualization and Export At the beginning of the project we decided that we wanted to provide some sort of graphical output for statistic results Since creating such functionality from scratch was not in the scope of our project we searched for open source solutions and found JFreeChart JFreeChart is a chart building library licenced under the GNU Lesser General Public Licence LGPL We quickly discovered that the best way to use JFreeChart within our frame work was as previously mentioned to provide some sort of wizard where a user can specify what exactly he wants to plot and how To create an com pletely generic solution where charts are generated automatically based on the result sets is although interesting to accomplish beyond the scope of our project This is why we decided to limit chart drawing functionalities to three basic but commonly used chart types Pie Bar and Line Charts A user should nevertheless be able to create more elaborate charts with statistical results thus why we decided that our framework also needed to have export functionalities namely to Comma Separated Value CSV files as they can be read by most spreadsheet software solutions such as OpenOffice Calc or Microsoft Excel With such software users can refine the presentation of statistical results and further processing the data Inttp www jfree org jfreechart
82. this area This is why we also contributed to Elexis itself and its user interface Figure 4 14 shows an example screen of Elexis in version 1 3 prior to our contribution There are many different styles of icons used throughout the application There is an additional inconsistency in the way forms look Some form elements have bad looking additional backgrounds Frames and borders are used in strange ways etc Datei Bearbeiten Fenster Hilfe AS ABE ARE Var CEP 50 2 E Patienten bersicht O Heute 8 Patie 22 ER Konsu H Labor Rp Rezep A AUF O Konsulta 33 Konsultation al d gt Arriba Samis 2 11 05 2007 nicht von A v heute test OH orname s Name Geschlecht Mobil Schon seid ich denken Absolut Erfunden w 18 11 1933 a kann war es mein Arriba Samis m 09 10 1981 sehnlichster Wunsch eir j Beeblebrox Trill 24 05 1978 Vorname Telefon 1 Fax F lltext zu werden Nun Briefe j SEH uh N Neco 1 rf ist es vollbracht Ich bin d Busch Dschordsch Dabbeljuh m 07 voller Stolz und 3 Cicero Marcus Tullius m 05 06 191 m chte mich am D Artagnan Aristide w 01 02 197C Geburtsdatum Telefon 2 E Mail GE lieder EIS ran yA erf gung stehenden Leistungen O Duck Dagobert m 09 12 1911 L cke breit machen Nicht jeder Text kann Hotzenplotz R uber m 03 04 1
83. ughout the book McAf07 This process was crucial for making some of 13 GO Sanclipse Prototype e 1831818 D Eleonora Luginb hl Female amp Hans Tkac Male 53 A Xena Siska Female N amp Markus Demenis Male F orename Hans amp Milan Tkac Male Surname PA amp Hans Tkac Male Sex Male D Eleonora Lindenmayer Female D Xena Siska Female D Jana Luginb hl Female D Martina Luginb hl Female amp Markus Schneider Male D Martina M ller Female amp Paul Luginb hl Male D Franziska Maier Female amp Markus Miller Male A Johan Tosi Male amp Oscar Horvatova Male amp Dennis Tosi Male amp John Mayer Male Sanclipse is a simple EMR system with the capability of exporting patients and their treatments in a standardized format OOO Add patient Please enter the patient details Sex Male sl Forename John Surname Mayer Cancel C Plug in Details 3 Configuration Details 4 Figure 3 1 Sanclipse Prototype Overview the architectural decisions described in the following sections and developing Archie 3 2 Ideas for Architecture The Sanclipse prototype described in the previous section is a stand alone Eclipse RCP application It is packaged and deployed on its own thus not depending on any other Eclipse program or framework Because Elexis itself already is a stand alone Eclipse RCP application
84. ults of a statistics are shown It opens a new chart creation wizard where users can build graphical charts based on the data in the results table ExportAction This action triggers an export dialog where the contents of the currently shown results table can be exported into a Comma Separated Values CSV file CreateChartsAction This action is used in the dashboard view to start the dashboard chart generation We implemented this action so that the chart creation could be triggered by the user instead of being started automatically as soon as the view is displayed This is particularly important for users with slower computers where the creation of these charts can take a while RefreshChartsAction This action refreshes the already created charts which is mostly needed when a user has been working some time since he generated the dashboard charts and needs to refresh them with up dated data 4 4 Managing Classes During the application runtime there are situations when we want to have only one instance of a certain object in order to either reuse it or its attributes or to be able to access it from different parts of the program while ensuring that the instance we access is the right one for all accessors This is why we implemented three manager classes An UML representation of all manager classes can be seen in Figure 4 8 Every manager class is an implementation of the Singleton Pattern and stores exactly one instance of a specifi
85. ype to the list of available widget types Totals in Results Table The result table in Archie is a visual representation of a dataset A dataset is modelled as a table as described in Section 4 2 and shown in Figure 4 3 Despite that there are several situations where a user is interested in the totals of columns currently there is no option to display the totals of the columns or rows of such a dataset in the results table A feature request based on this limitation is also listed among the list of issues on the Google Code project website Interaction with Results Table The results table in Archie shows different types of objects The dataset displayed in the table can hold simple types such as numbers or strings but also more complex and Elexis specific types like persons invoices or similar Currently there is no built in mechanism in Archie to allow contributing plug ins to show context specific information 2http code google com p archie issues list 49 based on an object type in the results table cell When a user for example double clicks a cell showing the value of an invoice in Elexis a contributing plug in might want to react to this event and show the details of the invoice in a separate window This limitation is also noted in the form of a feature request on the Google Code project website 4 11 2 External Internationalization Archie can be completely translated into different lan guages as well as almost every pa
86. ystem CIS is a clinical repository of patient data CIS is a form of EMR but it is more specialized to the environment of clinics Sometimes the term is used interchangeably with EMR A CIS typically covers pathology and radiology order entry and results reporting medication prescription and administration clinical work lists decision sup port etc Figure 2 2 shows a sample relation between these concepts Different EMRs respectively CIS are plugged into an EHR which accompanies a patient over his whole life Clinic Z Doctor X Doctor Y Birth Death Time Figure 2 2 EMR EHR CIS in sample relation 2 2 Status Quo The domain of electronic medical records is still pretty young and because of this standards one of the most important aspects of any such domain are still underdeveloped There are only a few accepted and implemented standards Also internationally and nationally acknowledged institutions managing and asserting such standards are sparse There are however some standards which are already well established like HL7 Health Level 7 a group of standards concerning the exchange of data between medical insti tutions and their information systems However there are also competing standards such as IHE Integrating the Healthcare Enterprise a European initiative with similar goals There are numerous other examples of conflict ing standards http www hl7 org http www ihe

Archie - Software Composition Group

Contents

Download Pdf Manuals

Related Search

Related Contents