מירב מימון טייב

We have previously developed an architecture and a set of tools called the Digital electronic Guideline Library (DeGeL), which includes a web-based tool for structuring (marking-up) free-text clinical guidelines (GLs), namely, the URUZ Mark-up tool. In this study, we developed and evaluated a methodology and a tool for a mark-up-based specification and assessment of the quality of that specification, of procedural and declarative knowledge in clinical GLs. The methodology includes all necessary activities before, during and after the mark-up process, and supports specification and conversion of the GL's free-text representation through semi-structured and semi-formal representations into a machine comprehensible representation. For the evaluation of this methodology, three GLs from different medical disciplines were selected. For each GL, as an indispensable step, an ontology-specific consensus was created, determined by a group of expert physicians and knowledge engineers, based on GL source. For each GL, two mark-ups in a chosen GL ontology (Asbru) were created by a distinct clinical editor; each of the clinical editors created a semi-formal mark-up of the GL using the URUZ tool. To evaluate each mark-up, a gold standard mark-up was created by collaboration of physician and knowledge engineer, and a specialized mark-up-evaluation tool was developed, which enables assessment of completeness, as well as of syntactic and semantic correctness of the mark-up. Subjective and objective measures were defined for qualitative and quantitative evaluation of the correctness (soundness) and completeness of the marked-up knowledge, with encouraging results.