III. Create Structured Rules
Once the guideline has been “marked-up”, GEM provides tools that work with the GEM XML file to iteratively refine the information, highlighting conditional logic, decision variables and actions. Through these steps, guideline implementers create structured logic specifications with which to implement the recommendations in the implementer’s EMR systems.
Steps
Frequently Asked Questions
- I. GEM-COGS Transform
- To facilitate appraisal of guideline quality, a GEM-COGS transform is run
- This shows the 18 criteria defined by the Conference on Guideline Standardization followed by relevant guideline text
- II. Apply Extractor Transforms
- XSLT EXTRACTOR transforms are applied to the marked-up guideline to extract implementation-critical information
- This creates a list of decision variables and actions for each recommendation.
- III. Adjust Level of Abstraction
- When “extracted” from context, it often becomes clearer which decision variables are vague, underspecified, or ambiguous.
- IV. Restate in Human Readable Logic
- Each recommendation is then restated in human-readable statement logic that can be translated readily into computable statements.
Guidelines
Organizing by Recommendation
Perhaps most valuable is a view of guideline knowledge components organized by recommendation and showing title, guideline narrative text, an IF…THEN statement of the guideline recommendation with each condition (decision variable) and each action parsed. In addition, fields are available to display the guideline text that justifies the recommendation (reason), the aggregate evidence quality supporting the recommendation and the guideline authors' view of the recommendation strength. This semi-formal view of the recommendation is ready for coding of the recommendation components into selected controlled vocabularies.
Visit this section to download examples from NHLBI, AAP, ASCO, and AUA.