2017-09-12 IA Small Group Call

Date

Attendees

Not Present

  • Claude Nanjo (attending HL7 meeting)
  • Joey Coyle (attending HL7 meeting)

Scribe

Goals

  • Review current ANF modeling guidelines for CDS KNART terminology artifacts

Discussion items

TimeItemWhoNotes
10minCurrent Status and UpdatesGroup
  • Kirsten:  Has not met with Claude since our Friday meeting. She did send the revised document with Keith's comments.  She did create a revised version of the document and will send it, along with any comments from today's meeting.
  • Keith:  Sarita, one of the things we discussed when you attended the call was to develop a list of your concerns.  Sarita:  Yes, and she needs to review the current document.
  • Keith:  There are 4 top-level models:
    • 2 of the models relate to action
      • 1 is for request
      • 1 is for performance
    • 2 of the models relate to performance measurement
      • 1 is for measurement
      • 1 is for measurement goal
  • Keith:  The first step is to define what the ANF will look like.  The CIF might be more specific, such as for pharmacy.
30minReview of Current ANF Modeling GuidelinesKirsten
  • Kirsten
    • The first version of the document was a proposal on their side that focused on the action request and how to separate and define the topic (i.e. the "what") and instructions (i.e., the "how") of an action request.
    • This document explains how to model it.
    • A glossary was added to the end of the document.
    • There are numbered principles in the topic and instructions sections.
    • A few more examples were added, such as to illustrate what we mean for timing or frequency.
    • Revised the tables and changed the content of the top-level table.  There are now less types of instructions.  Others will be included as sub-classes to instructions.  It now includes only instructions applicable to all types of action requests and none specific to a particular type of action request, such as route of administration that applies only to certain actions such as for medications.  Now it would be included as a sub-class for technique. 
  • Sarita:  Consult requests don't usually include a procedure site in the topic.  Keith:  It would probably be omitted.  It is the plausibility test.  You can say I want an orthopedic consult or one that relates to a patient's ankle, in which case procedure site would apply.  We don't want to talk about the middle of the road only.  We need to give some examples of the edge cases, like the one Sarita raised.  If there is an anatomic location involved, we need to clearly separate whether it's part of the topic or instructions.  We have to solve all of the problems in a model based on the use case.  Kirsten:  Maybe where there is an anatomical location, it will be included as part of the topic, such as "Head CT."  Keith:  When we pick something that is standard use in one area, for consistency we should apply in other areas.  When we take it to other areas, it may seem odd but we need to come up with the best compromise to apply consistently across all domains.
  • Kirsten:  Regarding separating instructions from the topic for a clinical statement, she showed the sub-specifications for technique, which would include the following:
    • Access device
    • Approach/access route
    • Route of administration
    • Projections
    • Device used
    • Device settings
  • Keith:  We want Instructions to have a general model that can incorporate domain-specific things such as projections, devices, approach/access route, etc.  The intent is there is a top mode.  For instructions, we need to insert the generic model and then indicate how sub-instructions would vary, such as for medications and would include route of administration.  He is appealing that what you have listed under Instructions are examples of what should be representable.  You have the top-level of technique, followed by category of Instructions.  We need a different name for the heading of Instructions, possibly Domain-Specific Details of Technique.  Sarita:  For the other types of Instructions, such as priority or precondition, will they each have their own sub-instructions?  Keith:  Yes.
  • Kirsten:  There were some examples in the document that we can further specify now, such as for medications or imaging orders.
  • Kirsten:  The glossary terms should be reviewed to see if everyone agrees with them.
10minDemo of Using Survey Monkey to Survey ANF Modeling ApproachesStephanie
  • Stephanie:  Showed the brief work she has done using the free online tool, Survey Monkey, to create a survey to collect people's input for how they would separate the portion of a clinical statement that represents topic and which part represents the instructions.
  • Catherine:  The University of Utah has a survey tool that is called REDCap that we could use instead of Survey Monkey.  Stephanie would create and administer the surveys but anyone could access them.  Catherine will check into it.  It's a very usable tool.
  • Keith:  He is trying to get reproducibility and validation of the models, both for CIF and ANF.  These may be some tools we can use to validate the model and write something up that says here is methodology we can use.
10minOpen DiscussionGroup
  • John:  Mentioned the need to create a use case for laparoscopic vs. open procedure.  Keith:  This is one of those edge cases he was describing earlier.  Please note this in your document for a procedure - does the open aspect of the procedure belong in the topic, technique, instrument, etc.?
  • Catherine:  She is reviewing the clinical white papers for the CDS KNART contract and is getting ideas for things that will need to be modeled, such as for surgical history.  Are we modeling those?  Keith:  We need to turn those into use cases and we're getting close.  We can't consume just a white paper.  We have to pull out the use case and the clinical data elements that need to be represented and then represent them.  Maybe you can bring some of those to the call on Friday.  If the use cases are bringing up a category that doesn't fit into the 4 models we have defined thus far, we need to know about it.