CPGPro – Extracting Process Information from Clinical Practice Guidelines

Like in many other domains, computerization is seen as a means to facilitate the effective use of Clinical Practice Gudelines (CPGs). Therefore, transforming CPGs in a machine-readable and executable format has been the gist of research. Consequently, many guideline representation languages are developed and modelling tools are designed which convert CPGs in their corresponding models defined in these guideline representation languages. Available modelling tools are restricted in the way that they only allow a manual computerization process. This feature makes generation of computerized medical guidelines a cumbersome task. Therefore, there is a need for automation of this process.

Along with my master thesis, an Information Extraction framework (CPGPro) based on knowledge engineering approach is implemented. CPGPro supports the automation of the mentioned computerization process. It automatically extracts relevant clinical actions and relations among them from otolaryngology guidelines and fills an XML template with the extracted information for subsequent processing by modelling tools. Processing is done in subsequent stages based on handcrafted lexical resources and extraction rules, which are implemented as heuristics derived from linguistic patterns encountered in otolaryngology guidelines. The documents of interest are XHTML-conform. The java framework, the otolaryngology lexicon, and the output template are available in the downloads section. An evaluation of CPGPro can be found in my master thesis.