Stream: implementers
Topic: Getting started with FHIR
Quin'darius Lyles-Woods (Apr 23 2020 at 06:07):
I have been reading 'Health Informatics on FHIR: How HL7’s New API is Transforming Healthcare' by mark l b. and I am in love with the field but can anyone give me some helpful clues on where to start with FHIR? even a good link or a book will be awesome you don't have to spoon-feed it. Anything is appreciated.
Quin'darius Lyles-Woods (Apr 23 2020 at 06:09):
and if this is a duplicate topic very sorry just link me to the main :)
Matthijs van der Wielen (Apr 23 2020 at 07:28):
I've added the book you have read to my list. Thanks! Unfortunately I don't have a book I can recommend for you, however I started with FHIR by walking through the Profiling Academy on https://simplifier.net/guide/profilingacademy/home, this really helped me to understand how to create profiles etc.
Lloyd McKenzie (Apr 23 2020 at 14:44):
Welcome to the community :)
The specification itself (http://hl7.org/fhir) is designed to be relatively approachable. At the bottom of the spec you can find a list of public test servers to experiment with. When you want to really dig into a topic, if you search for FHIR DevDays Videos, you'll find an immense set of short presentations on all sorts of topics.
...Paul (Apr 23 2020 at 14:55):
Is there a "Reference Implementation" of the FHIR schema?
Lloyd McKenzie (Apr 23 2020 at 14:56):
There are numerous reference implementations. Click on the 'downloads' link on the home page
...Paul (Apr 23 2020 at 15:03):
I saw those, but I guess I was thinking more of a database schema-level reference. The software downloads notes, that "implementations are not required to conform to these, nor are they subject to the formal standards process" which kinda contradicts the notion that they are "reference implementations". :D
...Paul (Apr 23 2020 at 15:05):
When I asked about Observations being "reused" by multiple DiagnostiReports last night, I was told the standard was "specifically vague" which makes me wonder how one should validate that their data model is, in fact, "standard".
Quin'darius Lyles-Woods (Apr 23 2020 at 15:06):
That's great advice! and yes the book is a fun read about the healthcare industry. I read Priced Out by Uwe Reinhardt and thought there has to be a solution to this and so far this book is trying to show one.
Quin'darius Lyles-Woods (Apr 23 2020 at 15:07):
@Lloyd McKenzie Awesome thanks Lloyd start in the test servers! got it!
Michele Mottini (Apr 23 2020 at 15:14):
You do not validate data models - you map (or try to...) whatever data model you have to / from the standard
In other words: this is a standard to transfer data between different systems - with different data models
...Paul (Apr 23 2020 at 15:23):
Okay, I guess that's what I'm missing/confused by. If I'm implementing an EHR backing store, should I not try to hew closely to the data model represented by this schema? If I make a design decision not to re-use Observations, for example... Won't that mess me up if I'm trying to import/maintain/transform data from another EHR that does? I thought this schema was intended as a common data model...
Michele Mottini (Apr 23 2020 at 15:26):
No, it is not a data model for implementation, it is a data model for transferring data - as such (for example) it tends to have more fields than working data models.
If you are implementing a EHR backing store use whatever data model makes sense for the EHR functionalities you need.
...Paul (Apr 23 2020 at 15:27):
What if I'm implementing a backing store and I want it to work with ALL THE MODELS? :D
Michele Mottini (Apr 23 2020 at 15:29):
Store the raw FHIR JSON in a JSON store...
Michele Mottini (Apr 23 2020 at 15:29):
That's what the reference implementations do more or less
...Paul (Apr 23 2020 at 15:31):
That's kind of what I was afraid of. :grinning_face_with_smiling_eyes:
Vassil Peytchev (Apr 23 2020 at 15:56):
...Paul said:
Is there a "Reference Implementation" of the FHIR schema?
What do you mean by "the FHIR schema"? What is listed here?
Lloyd McKenzie (Apr 23 2020 at 16:02):
There are a variety of persistence approaches. You can store using relational databases or no-SQL stores. If using relational, there are a variety of strategies depending on how you want to use/analyze your data
...Paul (Apr 23 2020 at 16:33):
Yeah, I guess I was looking for that level of "definition". Like, in my other question from last night, FHIR doesn't seem to define whether or not an Observation can be referenced by multiple DiagnosticReports or not (ie. is DiagnosticReport -> Observation a "one to many" or "many to many" relationship). But that level of specificity isn't really in the FHIR spec (and apparently on purpose). That just makes implementing a 'generic' EHR store tricky to assess as "correct".
Lloyd McKenzie (Apr 23 2020 at 20:54):
In general, FHIR tries hard to not impose too many rules - particularly around business process. That's because FHIR has to support a super wide range of processes in a variety of environments all over the world - and should ideally accommodate processes that haven't even been invented yet.
Lloyd McKenzie (Apr 23 2020 at 20:55):
The core FHIR spec is a poor source for 'correct' - you'll want to look at implementation guides in your country(ies) and discipline(s) of interest - and even those won't necessarily tighten everything down as much as you might wish.
Lloyd McKenzie (Apr 23 2020 at 20:56):
As a rule, if the core spec doesn't prohibit it, then it's allowed - though sometimes prohibitions might be documented in places other than the first place you go to look - so asking here and/or running sample instances through a thorough validator are always good things to do
...Paul (Apr 24 2020 at 15:16):
:+1:
JP Ventura (May 15 2020 at 10:47):
Is there a getting started or FHRI for dummies at the official documentation?
JP Ventura (May 15 2020 at 10:48):
Right now I am reading the https://www.hl7.org/fhir/ standard, but I have no idea how to create a wrapper of a legacy server compliant to HL7 v3 (authentication module only)
JP Ventura (May 15 2020 at 10:50):
Also is the documentation website content generated automatically from Confluence or is it saved in some https://github.com/HL7 public repository?
JP Ventura (May 15 2020 at 11:00):
Thank you!! Found it at https://github.com/HL7/fhir
JP Ventura (May 15 2020 at 11:01):
Now I need to find the specification of each entity at the repository
Lloyd McKenzie (May 15 2020 at 14:16):
The resources are defined in Excel spreadsheets saved as XML in the 'source' folder
Lloyd McKenzie (May 15 2020 at 14:17):
We publish using a (very sophisticated) home-grown Java app
Last updated: Apr 12 2022 at 19:14 UTC
