Stream: implementers
Topic: New resources recommendation
Alexander Henket (Aug 14 2017 at 08:25):
We're facing a large body of work that doesn't translate well into existing resources. So we could
-Create a Basic resource for every chunk
-Create new resources
Either solution could be followed up with a proposal to add them to the core. Considering that it evolves around use cases that are presumably too Dutch to make international, we might not get those approved.
Are there any recommendations to make here?
Michel Rutten (Aug 14 2017 at 09:10):
Hi @Alexander Henket, if any of the proposed new resources won't reach approval, then they probably also won't be supported by any available FHIR tooling (out of the box).
Alexander Henket (Aug 14 2017 at 09:12):
True, but they wouldn't be supported at the very start either. We will build stuff for STU3. It's not until R4 they we _may_ get out-of-the-box tooling support.
The tradeoff is that while Basic resource profiling leverages built in FHIR capabilities, it is a “dragon” compared to the concise (domain specific) XML it is today. Building new resources however is even worse because it is very possible that what we create is never accepted internationally.
Grahame Grieve (Aug 14 2017 at 09:26):
Basic is not much fun. It'll get you started, maybe. If has less risk - a little less. The best way is to describe the scope of the things you talking about here, and have discussion on the likely scope here, then run by the likely committees, and seek for contributers
Grahame Grieve (Aug 14 2017 at 09:26):
then design your own, given the comments you get, and then take back to committee
Alexander Henket (Aug 14 2017 at 11:45):
I agree that's the "royal way" (Dutch expression). When it comes down to creating my own set: hoew does that work? Can I just start a StructureDefinition on a new type of resource? Do I start from Excel? Do I start from a logical model?
Grahame Grieve (Aug 14 2017 at 11:47):
we're trying to move from spreadsheets for profiles - and it's not working very well. I'm not even trying for resources. It's spreadsheets only.
Grahame Grieve (Aug 14 2017 at 11:48):
you can start with a logical model, sure. but eventually it has to become a spreadsheet representation if it's going to be a real resource. Though it probably wouldn't be to hard to write a converter
Alexander Henket (Aug 14 2017 at 11:48):
Alright. That just limits my (initial) processing options to the publisher I guess — we normally work through Simplifier
Grahame Grieve (Aug 14 2017 at 11:49):
maybe. slow build though. I might even prototype with a logical model myself now
Alexander Henket (Aug 14 2017 at 11:50):
Slow is better than not at all :-) Logical Model is definitely an option although we have none so far
Lloyd McKenzie (Aug 14 2017 at 14:00):
Keep in mind that if you use custom resources, you can't claim to be FHIR conformant. It would really be helpful to get a sense of what sort of resources you're expecting to need
Michel Rutten (Aug 14 2017 at 15:42):
Hi @Alexander Henket, if desired, we could implement support in Forge for authoring (new) core resources/datatypes. Seems similar to authoring logical models, which is already supported, so it shouldn't be too hard.
However support for custom datatype constraints on element types is a bit trickier to implement...
Alexander Henket (Aug 15 2017 at 06:47):
@Lloyd McKenzie: understood. The domain is too much to cover in whole here, but I'll try to give a taste:
https://www.istandaarden.nl -- This is a Dutch language website that deals, among other aspects, with communication patterns around 4 bills (laws). Long term care, social care, youth care, mental care. They are mostly administrative in nature. They for example deal with whether or not a person is eligible to receive care/help coordinated by their municipality. The types of services they could be eligible for vary from vacuum cleaning, washing, to changing of "supporting stockings".
The first use case we're going work on is the decision report that details their eligibility for these municipal services. The information model for that is here: https://modellen.istandaarden.nl/wlz/iwlz1_2/index.php/IO31_(View). They now have their own XML format for sending stuff.
What you'll see in Dutch:
- Header == FHIR MessageHeader [we'll skip that]
- Client == FHIR Patient
- Relatie/Contact == FHIR RelatedPerson
- Indicatie == ... [this is where the Indication decision is]
- GeïndiceerdZorgzwaartepakket == ... [a package with a certain weight. More weight is eligible to more care]
- GeïndiceerdFunctie == ... [a physical or mental function that gets attention under the decision]
- Stoornis == FHIR Problem [background for the decision]
- StoornisScore == FHIR QuestionnaireResponse [background for the decision. questions related to the Stoornis]
- Beperking == FHIR Observation? [physical or mental impairment/limitation]
- BeperkingScore == FHIR QuestionnaireResponse [background for the decision. questions related to the Beperking]
The Indicatie [Indicated decision on care eligibility] sort of resembles EligibilityResponse but has very different properties. It also has some characteristics of a Contract, in the sense that you can hold your municipality to this decision when it comes to reimbursement.
For the truly curious: the full list of messages they currently have is here. To the right you may download their XSDs for that: https://modellen.istandaarden.nl/wlz/iwlz1_2/index.php/Berichten
Alexander Henket (Aug 15 2017 at 06:51):
@Michel Rutten That is a tempting offer. Would the FHIR Community itself not benefit from creating reusable patterns, i.e. datatypes through Forge, regardless of my particular use case? For our purposes I first need a little more work on our attack strategy before asking new features from Forge. Thanks though, I'll keep it in mind
Grahame Grieve (Aug 15 2017 at 06:52):
@Bryn Rhodes a lot of this stuff sounds like decision support outcomes
Grahame Grieve (Aug 15 2017 at 07:04):
@Alexander Henket will you be in San Diego? At the connectathon?
Alexander Henket (Aug 15 2017 at 07:05):
No sorry. Funding thing. I'll not be in San Diego at all
Grahame Grieve (Aug 15 2017 at 07:05):
these kind of questions require funding... well, we'll pursue them here
Alexander Henket (Aug 15 2017 at 07:06):
Well there's funding for doing work on this, but that does not include going to the WGM this round :(
Lloyd McKenzie (Aug 15 2017 at 13:54):
The contractual aspect sounds more like a pre-determination than an eligibility check. In the former, you're presumed to have some sort of coverage and the question is "if we were to provide services of this type, would the cost be covered (and possibly how much of the cost would be covered?". Eligibility is more about "if you were to seek coverage, would we likely give you coverage?". My leaning would definitely be to use the existing resources, generifying them as necessary, rather than creating something country-specific. The scope you appear to be talking about doesn't seem like something that's country-specific - though of course some of the data-elements might be.
Bryn Rhodes (Aug 22 2017 at 15:50):
Definitely agree these sound like decision support outcomes, and that the current resources should be able to cover the use cases here.
Brian Postlethwaite (Aug 27 2017 at 09:16):
Have you also looked into the EpisodeOfCare resource also for the tracking of things?
As activities occur I expect that some will be allocated to the various types, and EpisodeOfCare can be used to track this.
Last updated: Apr 12 2022 at 19:14 UTC
