A Diabetes Data Browser Using Java and Semantic Web

This thesis documents the creation of a semantic data browser prototype for RDF (Semantic Web) data primarily intended for documentation and follow-up of diabetes. The application includes routines for importing data from glucose meters (glucometers) through a serial port. A graphical data browser showing tables and time-based diagrams was created. All data items are clickable in order to allow editing and to retrieve more detailed information than what is available in the condensed graphical overview. The browser could also be used for other medical or non-medical purposes but that has not been tested. The browser is platform-independent and written in Java. RDF was chosen for internal data modelling. RDF is also used for export and import of data. Email and file import/export of RDF models/data is provided. The implementation uses the “Jena” toolkit from HP for RDF processing. The browser is intended to display and modify instances of RDF data according to user supplied schemas and settings; it is not designed for schema editing (there are better tools available for that). The application can be partly extended simply by providing more knowledge about the world through RDF-based schemas/ontologies. Settings and some user interface configuration are also adjusted by editing RDF files. Other parts of the application require Java programming in order to be extended, prepared extension points are documented. The whole application is available for free as Open Source to allow full modification and free usage for anyone. Background chapters in the thesis briefly cover Semantic Web technology, data modelling, system environment, diabetes and healthcare requirements. Suggestions for improvements and future work are discussed. Dedication This thesis is dedicated... • ...to my wonderful wife Maria and to my son Benjamin, thanks for everything and for giving me a lot of time to finally finish this thesis. • ...to my mother who died in cancer before the thesis was finished, and to Clarence and the wonderful home healthcare staff that in a professional and loving way helped her during the last years. I hope the results of this thesis sometime in some form will be useful for home-based healthcare in the future1 Acknowledgements This thesis has been made possible by the help of many others, special thanks go to: • The staff at the Dept. of Biomedical Engineering (IMT) at Linköping University, especially Hans Åhlfeldt, Leili Lind and Nosrat Shahsavar in the Medical Informatics group • Santa Anna IT Research Institute, www.santaanna.se (in cooperation with SITI www.siti.se) especially Sture Hägglund • Lars Olsson at Cendio Systems, thank you for supporting the start of the thesis financially. • Bayer, Medisense, and Roche thank you for supplying glucometers to the project for free. • Ann Björklund and her colleagues at the diabetes clinic • The “medpum” student group (See chapter 5 and [medpum 01].) • Sourceforge – Thanks for supplying CVS-services, hosting project web pages, mailing lists, download servers etc. www.sourceforge.net • OpenOffice.org The thesis was written and illustrated mainly using the word processor and other tools freely available as Open Source from www.openoffice.org • Programmers behind utility classes and packages used, see chapter 7 for details. • The Jena team at HP Labs, thank you for making RDF processing simple in Java! • The teams behind the great development tools IntelliJ IDEA, www.intellij.com and Eclipse, www.eclipse.org Errata and updates If corrections of errors in the text and/or other thesis related updates are published (after this thesis goes to print) they will be available at http://ehealth.sf.net/thesis2003/. 1 If something in this thesis gets used in home-based care I'd be grateful if you mail me a few lines and describe how. My private email (spam-protected and unlikely to change) is: [email protected] Table of