20min | Update of Current Status | Group | - Keith: Would like for Stephanie to participate in the Terminology rounds, and in particular to pull all of the documentation together, which has been sort of elusive in the past. The work she has done in the past has been very helpful in keeping it going. We have the ONC Learning Series in January, and he would like to try to have the ANF modeling guidelines ready by then. He also thinks that Kirsten has a style/editorial guide for LEGO modeling that may be suitable for contribution. Kirsten: She isn't sure how beneficial it would be but she is happy to provide it. Keith: We know everyone is pressed thin and different priorities can affect someone's dependencies, and often, documentation is the last thing that is done. Having Stephanie help with this will ensure we get the documentation done. Stephanie: She'll need to know which calls she needs to attend and have someone send her the invites, if she is not already attending them.
- Claude: For a quick update, Joey has been working on the revised TSR. Today we had a meeting with Sarita and her group to review it as they begin using it to model. Next week, they will apply the ANF model on a very simple, bare-bone documentation template in order to test out the TSR and the model. After that, we'll get back to you next week. We want to get this into full motion. Keith: Any meeting with Deloitte and relating to KNARTs that Stephanie be invited to. Stephanie will need to manage her calendar and determine priorities but she should be at least be given the opportunity to join the meeting. His schedule is overwhelmed. He would like to have the Small Group meet once a week. Sarita and Kirsten, does that make sense to you to bring in Stephanie? Sarita: Yes, and they have one call a week and they'll be sure to include her. Stephanie: Currently we have the Small Group calls twice a week, once on Tuesday and once on Friday. Which one are we cancelling? It was decided to keep the call scheduled for Tuesday and cancel the Friday call. Keith: During Thanksgiving week next week, he is available Monday-Wednesday but we might want to cancel next week's call. We will meet the following Tuesday. It was agreed to use Tuesday's IA Small Group timeslot instead to discuss modeling of KNART elements with the Terminology team, provided there are no conflicts.
|
30min | Review of Current TSR Spreadsheet | Group | - Keith: One of the things we didn't get an emphasis on is figuring out what the categories are. We don't want to ignore the issue and not do a good job. We've had discussions about techniques and preconditions and how to figure out the differences between the two. He has had discussions with Joey and it may be a false dichotomy. We need to come up with a set of categories that we can separate them into but we don't want to invent false things, such as we have now with finding. If we can't come up with reproducibile distinctions between them, then it's likely they should be merged. We had a similar issue with action performance vs. phenomenon measurement, such as patient is sitting during BP measurement. It got really convoluted to keep a distinction that may not be valid. Essentially an action may result in a measurement and the other is a request. There are still challenges to work through, such as how do you know that an action doesn't result in a value and should every action have some value that says the performance was successful, which isn't necessarily optimal but was performed to completion. These are issues we need to go through. As you're filling out the details, pay some attention to what are the categories and can we provide a textual definition of what it means to be included and excluded from the category. We can expect that there will be monthly updates to this spreadsheet. HIs goal for each monthly cycle you have a goal for the types of things. But there is a dependency on the CDS team and what they want to push. But, he is depending on Claude to make sure it's stacked and we're not just focusing on one thing, such as just orders, and it needs to be a representative sample. Claude: It won't be just one type of thing. We have 104 KNARTs spread across many different types of domains. We'll have a solid cross-section. Keith: An important deliverable at the end of each month is a style/editorial guide that says how you applied the style sheets and what worked and what did not work. This is where he said Stephanie will become involved.
- Keith: Where we have detail.value.code, we have to figure out the key. He assumes that a patient "is sitting" during a BP would be an example of what would be included here. There are different ways to construct these keys and codes so that no two follow a common pattern. The focus has to be on finding reproducible patterns for these things and then categorize these things and have particular patterns for them, such as a pattern for technique and a pattern for precondition. If things don't fit, then we need to determine what needs to be changed, such as the editorial/style guide. He hopes all of this makes sense. He's been trying to be consistent on this for a while. How can he help make the task more clear, are there lingering questions about how to approach this? Claude: He defers to the Terminology team because the work of how to structure the content in each of these slots really is a Terminology decision. But, we need to have clear guidelines and principles. Keith: Agreed. We must have clarity. We're not interested so much in correctness. We're not objecting to a principled approach that may emerge from somewhere. But the fact that two people may have valid differences, at some point we have to just pick one and determine which is more consistent with the higher level principle. Kirsten, does this make sense to you? Kirsten: Yes, reproducibility is the important thing and sometimes we do have to just pick one and say that's the way it is. Some things we'll need to try out. Keith: Is there any lack of clarity as to what the task is or that you need better understanding of before moving to the next steps? Joey: The topic type, which is the action, in the examples he included things like "imaging-of", if you look at "CT of the Head," he can do it two ways. One way is where device = CT or could have done it the way SNOMED does it. If the latter, CT-imaging of, then there isn't the detail of CT. He thinks we should make our topic types more general rather than bringing into the action. Keith: Let me make sure I understand the question. Consistency is your absolute most important principle. If following SNOMED makes it more inconsistent, that would be a bad choice. SNOMED may be for CIF. If we make ANF less consistent, the less usable it will be. Fir the topic type, that is something you'll need to work out. What is its definition? For it, we have two top-level things, and one is for Action and one for Request. For Action TSR, we are saying topic type must be a sub-type of action and it makes sense to use things like observation-of and imaging-of. He thinks observation-of and measurement-of are the same thing. Does that principle make sense? Joey: He understands it's basically that the principle must be to make consistency happen. Keith: Is everyone willing to accept that you may need to set aside the details of a particular model (missed part of what he said). We want the SNOMED model to inform our work but they are to inform our work and they are not dogma. They are just as likely to contain errors that prevent consistency in the ANF. We need to understand the issues and document them well. It may be that it needs to stay in CIF and not just say it's because SNOMED or LOINC does it this way. Claude: We should present to those groups when we believe there are inconsistencies that need to be fixed. Keith: Yes, but we're not seeking approval from these groups, including CIMI, for making our choices. We want these things to be use case-driven. For FHIR, measurement of absence comes to mind. We need to have a use case and we should think through a rational alternative to consider. Kirsten, does what he describe WRT the SNOMED model and how it should be considered make sense to you? Kirsten: Yes, it looks like we'll need to attempt with the attribute and qualifier values and the topic, domain, and others to use other terminologies (not sure I captured this correctly). She isn't sure how LOINC would fit for imaging. Keith: To the extent they have a well-articulated model for the model we want to try, that is fine. Sarita: She understands what he is saying and how they should proceed in this manner. Right now, it's open-ended as to the various possibilities. They'll need to find the modeling patterns and where they can find the consistency and move forward with a future model.
- Stephanie: Requests that she receive TSRs with the real information included. Keith: Yes. Please make sure Stephanie sees everything and include her on the meetings. Sarita: We will.
|