Standards Roadmap 2014 minutes
Attendees: Anisa, Bob, Gitte
- Discussion: minutes should remain circulated in emails, but should also be posted on Wiki.
o Action Item: Bob will post all 2014 minutes on Wiki by next Tcon
- Discussion that the group will maintain weekly meetings until the CSS PhUSE 2014 meeting in mid-March.
o Invite will be sent to reflect our weekly meeting on Outlook.
- For now, we’ll keep the 10 am (Eastern) Tuesday meetings and everyone can remain up-to-date via meeting minutes. If attendance suffers, we can modify our timing.
How to Design a Domain-
Decision Tree Discussion-
Discussed: Should mandatory variables be in a single bucket? Or can some of the mandatory variables be placed into other buckets with similarly-themed (non-mandatory) variables?
- Discussed the idea of pulling some Core Domain Variables out and placing them into other buckets. We developed some thinking (pros and cons) but will leave a decision till more people are on the call.
o Keeping the mandatory variables together keeps our project aimed at “designing a domain” (thus helping someone build domain). Keeping our variables in strict concept-based buckets focuses on categorizing variables (help people understand how variables exist in themes). The group should discuss this in the next call. Either approach could work.
- Anisa observed that some variables are mandatory, and thus aren’t really a decision. These should reside in one of two places: 1) up in the first swimlane, or 2) some of them may exist in the left-most part of their relevant swimlane, perhaps in a block at the beginning of the swimlane. o This discussion got us thinking of alternate ways of representing the decision-tree.
Bob wondered if the first swimlane remains as-is, then subsequent swimlanes are turned 90 degrees clockwise with the mandatory variables of a bucket located toward the top of the swimlane.
Gitte offered that a “wheel” might work… with mandatory variables in the center and each spoke (or slice) around the wheel is a different swimlane containing a set of similar variables.
• Bob and Gitte will make rough drafts (on paper) and scan to visualize the ideas for discussion at next meeting.
- Issues to resolve:
o Do mandatory variables belong in a bucket?
o Do we keep visual as-is? Or wheel? Or modified-swimlane (some rotated)
o Identify variables that do not belong in our buckets. Do they belong in a “miscellaneous” final swimlane? Do the get a “note” below the decision trees?
Note: A section in the accompanying text should address uniqueness of records and an explanation of natural keys. This was a spin-off from a discussion about the ‘Test Variables’ and the need to specify these further.
Tcon next week (1/28)
Meeting Minutes 1/14/14
Attendees: Donna, Gitte, Anisa, Bob, Lynda, Rick
1. The group discussed a new time for the Roadmap Tcon - 9am (Eastern) on Tuesdays fits most schedules, but not all. Come to the next meeting with an idea of: 1) Is 9am on Tuesday better for you? And 2) Is there any other desired times (casting a wide net here)
2. Bob distributed a Word doc the will evolve into a Purpose and Instructions document for the “How to Build a Custom Domain” project. It’s not finished by any stretch, but members are asked to read and provide feedback at this point. How detailed do we get on this doc? Do Use Cases belong in here? Instruction must be further filled out, particularly after we build consensus after our piloting exercise (below).
3. Gitte shared her (extensive) progress on the decision tree and domain table. Thanks Gitte!
a. The version attached to this message is the most recent version. Changes to the domain table are made in tracked changes.
Action Item for all members:
- Test the attached decision tree and domain table. If you have study data, you can use it in your test. Otherwise, you can do a “run though” of the decision tree without actual data. The key is to go through step-wise and assess our current work items. Come to next Tcon with comments/critiques on things such as (but not limited to) 1) Are the categories appropriately titled and in an intuitive sequence on the decision tree? Based on your usage, did the decision tree fit your thinking? When going down swimlanes, are the variables grouped appropriately? Which variables belong in our final “Other” swimlane (not currently shown on tree)?
o If you kept e-notes during your exercise, we can share during the next Tcon
o This is an example of building through piloting, so your work here is very much appreciated.
Roadmap Tcon - 1/7/14
Attendees: Anisa, Gitte, Donna, Bob, Rick and Debra
Discussion about Custom Domain Structure and Accompanying Visual
The group discussed how many “buckets” accommodate the variables in the domain structure. Bob shared his “bucket exercise for variables” and Gitte shared her draft of a visual. The visual is a series of swimlanes, each lane having a decision tree that ends in a subset of relevant variables . Subsequent discussion was on the number and nature of the buckets, and then how it meshes with the visual.
- It was determined that the buckets would be a “diving platform” for each swimlane in the visual. Users are expected to start at the top left of the document and move along the first swimlane before moving to the next lane. It is understood that, for some data types, a custom domain many not include info from every swimlane.
- Buckets (thus far) include: Mandatory Domain Variables, Test Variables (this bucket probably requires further discussion), Objects of study (composed of Subject ID, Specimen, Locations) variables, Result Variables, Timing Variables. There are variables that don’t fit into these buckets, so additional buckets (including Miscellaneous) will be considered.
o (thought after Tcon) During next Tcon, we’ll need volunteers:
Each member will be assigned a bucket and “claim” variables for their specific bucket (one bucket per member). This is a confirmatory exercise. In the end, each variable should only be in one bucket (no overlap).
a volunteer who will identify all variables that don’t fit into any of the existing buckets. These “left over” variables might be a “side order” on the menu for specific needs, and might be considerations to address in a final “swimlane”.
- The group felt the swimlane visual suited the goal and that the buckets listed above were applicable. In general, the group agreed to work their way down content for a single swimlane at a time.
- Some variables may belong in two or more boxes in a single swimlane. The group discussed whether these needed to be represented differently in the swimlane, but it was agreed that we’d need a visual before making a decision.
- Each member should continue thinking of the most intuitive way to represent the visual and table. Consider real study types and how its custom data might fit into this domain.
- Gitte agreed to continue work on the swimlanes and incorporate a color-coding to boxes that will correspond to color-coded buckets of variables in the domain table
- Bob agreed to start a document that states our purpose and instructions on how to use the visual/domain structure. An updated domain table will be circulated after the Tcon (coloring is tentative and subject to change as necessary).
-Next Tcon is 10 am next Tuesday, using our original Webconnect and Tcon number
Thanks to everyone for a great start!