Stream: implementers
Topic: Graph Definition - example
Kevin Mayfield (Feb 23 2019 at 08:57):
I'm exploring using GraphDefinition to describe a MessageDefinition. The example I'm using has one focus of Bundle in the MessageDefinition.
I'm not sure my GraphDefinition is correct, this is the gist I've done so far: https://gist.github.com/KevinMayfield/0714ee1a5785779eb389414d8285d365
I've also done some work on rendering this, top part is shown below:
Screenshot-2019-02-23-at-08.45.01.png
It seems correct but:
1. I don't have a profile for the Bundle but think I need a link back to the MessageDefinition (so this says this is a GraphDefinition for this MessageDefinition)
2. Have I interpreted the MessageDefinition correctly. I've taken the MessageDefinition as implying a Bundle with MessageHeader as the first resource.
The Bundle in this case is a FHIR Document which I started defining as:
Kevin Mayfield (Feb 23 2019 at 09:01):
Screenshot-2019-02-23-at-08.45.17.png
If I keep going with this FHIR Document, it's going to be a big GraphDefinition and wondering if would be better to refer to a separate GraphDefinition which is just the FHIR document. Which makes me think when I get a resource Type I would normally have a profile but for resourceType = 'Bundle' I may want to reference a MessageDefinition or GraphDefinition instead?
Kevin Mayfield (Feb 23 2019 at 09:03):
I'm targeting the rendering at developers (which is why I've focused on Bundle), I may use a graph chart for other users.
Zoltan Simon (Feb 23 2019 at 12:55):
Hello , I am new to this community and the domain.On the graph topic, have you considered using Neo4j and Cypher to describe this as a graph? It also can give you instant visualization and query capabilities through GraphQL e.g.
Lloyd McKenzie (Feb 23 2019 at 15:40):
It's probably reasonable to have a node in a GraphDefinition point to another GraphDefinition. I wouldn't focus on the other entries in the Bundle - what really matters is the path through MessageHeader to focus. We might need to bring the notion of ElementDefinition.type.aggregation into GraphDefinition too. Do you want to submit change requests?
Grahame Grieve (Feb 25 2019 at 01:13):
we already described how to use graphql with FHIR
Kevin Mayfield (Mar 03 2019 at 09:41):
I'm still looking into this. Looks like I can achieve my first aim with multiple GraphDefinitions on the MessageDefinition.
Last updated: Apr 12 2022 at 19:14 UTC
