Notes

Outline

Mark-up based analysis of narrative guidelines with the Stepper tool Marek RŮŽIČKA and Vojtěch SVÁTEK University of Economics, Prague (UEP) and EuroMISE Centre – Cardio, Prague

Structure of the presentation Step-by-step mark-up based formalisation of narrative guidelines Stepper tool UI & formalisation example Transformation process diagram Comparison of existing GL mark-up tools Future work

Step-by-step mark-up based formalisation of narrative guidelines (1) Bottom-up, document-centric approach to formalisation vs. top-down methods relying on a flowchart-like model Advantages of bottom-up approaches lower risk of information loss lower risk of implicite subjectivisation Disadvantages of bottom-up approaches might be tedious large syntactical and semantic gap between marked-up text and operational model

Step-by-step mark-up based formalisation of narrative guidelines (2) Central ideas of the Stepper tool: An ‘intelligent’ mark-up editor and transformation processor might make the formalisation easier Explicit separation of formalisation levels might help to bridge the mentioned gap

Step-by-step mark-up based formalisation of narrative guidelines (3) Stepper offers initial text mark-up rule-based transformation between different levels of formalisation or different models retrieval of corresponding knowledge elements (and text) across formalisation levels Used technology standards: XML, XLink, XSLT original XKBT (‘…knowledge base transformation’) language defining interactive parts of transformation

Step-by-step mark-up based formalisation of narrative guidelines (4) Experiments WHO 1999 Hypertension Guidelines tentative formalisation of selected parts through all levels of formalisation result: automatically generated demo-application Czech Unstable Angina Guidelines Testing formalisation of the whole document Breast Cancer Guidelines (at VU Amsterdam) preparation of ASBRU language implementation

Stepper tool UI (1) Initial mark-up of source document

Stepper tool UI (2) Typical work with Stepper : XML-to-XML transformation

Stepper tool UI (3) Example of creation of asbru plan from several parts in previous level Transformation rule types: Aggregation Decomposition One-to-one relation All transformation rules are prepared in included editor

Transformation process diagram (1) Multiple step approach disadvatages With increasing number of levels whole process is getting unbearably complicated UI for only two levels at the same time is insufficient Possible solution Each new succesor is enriched with links to all its ancestors (using Xlink technology) Links allows finding all text fragments in source document for any part of any level

Transformation process diagram (2)

Comparison of existing GL mark-up tools (1) Existing tools: GEM-Cutter Developed at YCMI Supports only GEM format Graphic Mark-Up Tool (GMT) Developed at VUT Supports Asbru language Uruz Part of DeGeL project Supports both GEM and Asbru (possible extension to other ontologies)

Comparison of existing GL mark-up tools (2)

Comparison of existing GL mark-up tools (3)

Future Work Co-operation with the database group at the EuroMISE: resources such as ICD-10 or ATC already part of their EHR application MUDR stand-alone and web-service-based provision of terminology for Stepper envisaged Including other ontologies, e.g.:  GLIF (version 3 is already based on XML and has execution engine GLEE)  Arden syntax (many documents encoded in MLM) More information at euromise.vse.cz/Stepper