2017-09-05 IA Small Group Call

Date

Attendees

Scribe

Goals

  • Review Claude's draft document for proposed ANF modeling guidelines and his draft spreadsheet to use to survey participants on how to model various clinical statements.

Discussion items

TimeItemWhoNotes
10minReview of Current Status with ANF ModelingGroup
  • Claude:  At our last Friday call, Kirsten had submitted a proposal for the separation of topic from instructions.  Our next step is to deploy on a smaller scale and have people try out the spreadsheet and then reach out to a broader group on the Deloitte and Terminology sides, as well as to VA SMEs and Cognitive SMEs.  We would then determine from their responses if our guidance on how to separate topic from instructions worked or not.  At the very least, it would have a set of principles for usage of the CIF and ANFs.
  • Stephanie:  How is this different from the current TSR that is being reviewed and for which feedback is currently being submitted?  Keith:  This is for the Action Request clinical statement whereas the previous one was for Phenomenon.  We want to have a process where we state how to get from clinical statements and to ANF.  He wants the whole process documented so it's like a recipe.  We want a clean separation of what is the part of the model that terminology populates and what is part of terminology (missed the rest of what he said).  Claude:  Now we are formally documenting what we have been discussing during several sessions.  Keith:  He would like some description of this as part of the upcoming SOLOR release.
 30minDocument Review Claude
  • Claude:  He will send the document to everyone included on this meeting invite.
  • Claude:  The first part is an introduction.
  • Keith:  He wants a set of numbered principles, which may form a decision matrix.  There is an elaboration of each principle and an example of how you might use it. Claude:  He will make this addition.
  • Keith:  We want symmetry.  Under Instructions, you have some things which in Phenomenon we are lumping into the Topic, such as a precondition.  It bothers him that we are not consistent.  We'll need to take what we're doing for Action Requests and apply them to Phenomenon.  Claude:  For measurements, it's driving to state (e.g., patient has X and here is the value for X) but for measurement, it's telling you how it was taken.  Those are really attributes of procedures and are instructions/methods for how the procedure was carried out.  It should be symmetrical.
  • Claude:  That was a general set of guidelines and then they are refined starting in Section 3, Modeling Medications.
  • John:  "Target" is very overloaded in this context.  Catherine:  For her, she was thinking of patient vs. donor, for example.  Keith:  The language will come and we need to keep noting these issues.
  • Keith:  We want the instructions for a request to have fields that are general across all of these categories.  How can we normalize instructions?  A frequency is not unique to a medication order.  Claude:  In section 2, it pointed out there are attributes that are unique to certain types of procedures.  He understands what Keith is saying and that there is a way to do it.  A problem can occur though when you generalize an attribute name as it can be difficult to figure out what it is in a particular context.  Keith:  We need to define what is an interface to instruction.  The next step is within the instruction, are there things we can pull out that are general, such as precondition?  For medications, the goal as he sees it is to map them in the same way as opposed to introducing new things, such as dosage.  You're letting concerns leak, which we need to try to get out.  Claude:  This may be a difference in object-oriented programming modeling vs. modeling of sets.  Keith:  A step for Friday's call is to take the general view you have in section 2 and in sections 306 and you explain that you are not allowed to introduce any new instruction types.  Rather, for example for a respiratory therapy order, you say here is how the aspects of it would apply.  Catherine:  It would help if you had a table that displays how each type of instruction applies to each type of action request.  Claude:  He won't be able to attend Friday's call as he will be attending the connect-a-thon but will work on the homework for Friday's call.  Kirsten can present if she is comfortable with the changes that need to be made.  We can then review it on Friday.  If there are any action items for him from that call, please let him know.
  • Keith:  The top level is the who, what, where, why, and when, and dosage could be part of the what.  
15minSpreadsheet ReviewClaude
  • Claude:  There is an Instructions tab.
  • Claude:  In the worksheet, we will have a set of examples that span a number of types of action requests, as shown starting in column B.  Reviewers can then indicate how they would model the topic and instructions and provide comments on how they did that.
    • Example:  Administer 325 mg chewable aspirin.  One might model is as the entire topic is "Administer 325 mg chewable aspirin" with no instructions.
      • Catherine:  Her gut reaction was to try to divide it up into the attributes you described in the other document.  
      • Keith:  If we have instructions, then we would breakout the separate categories and we will perform that exercise later.  For now, we just want to see how they would split the topic from the instructions.
5minNext StepsGroup
  • Keith:  Asked Claude to begin working on Phenomenon Measurement.  Keith wonders how different it will be from Action Performed.   The main thing is we need to have the symmetry between the two.  One is resulting in a state and one is resulting in a measurement. 
    • Example:  You went from having a state of having an appendix to not having an appendix.
  • Claude;  He will take a stab at the higher levels of the decision tree.  In the meantime, the focus will be on finishing Action Request changes.  Everything ultimately is procedure-centric.

Action items

  • @Claude:  Add a set of numbered principles to the document.
  • @Claude:  Make changes to sections 3-6 in the ANF modeling document for Friday's call.  Send it to Kirsten for review and then send the main document and spreadsheet to Catherine.
  • @Claude:  Begin working on Phenomenon Measurement.