Stream: implementers
Topic: Observation Patterns
Dave Barnet (Jan 21 2020 at 09:30):
I'm just starting to look at "doing observations". As this topic is so large and wide ranging, I thought that I might start looking into patterns of Observations. There are
- simple key value pairs (eye colour obs, clinical findings) that are perhaps a code and value
- vital signs (respiratory rate) that follow the FHIR pattern
- composite readings (BP) that use the component values
- Blood test results that use DiagnosticReport & Observation
- Wave forms measurement (not sure how to model these?)
- Genetic type observations that produce large amounts of sequencing data (MolecularSequence?)
- and others ..
So my initial thoughts were to describe the various patterns & how they are modelled in FHIR, then have a decision tree to help decide what pattern a particular Observation conforms to. Does this sound like a good idea, or am I over simplifying what is a complex area? Has anyone done any work in describing the types of Observations that occur in healthcare?
Domenico Corapi (Jan 21 2020 at 13:26):
@Dave Barnet have you looked into FHIR profiles?
https://www.hl7.org/fhir/profiling.html
Example:
https://simplifier.net/finnishphr/fiphr-waistcircumference-stu3
This one specifies a fixed value for category, code (loinc|8280-0 for waist circumference), and unit of measure (cm).
Dave Barnet (Jan 21 2020 at 15:59):
Thanks for the reply. So what I'm after (taking your example) is to have a decision tree that would decide what observation pattern an observation would conform to, and what shape that observation would be. So for the waist circumference example you gave, that is declared as being a vital sign key-value pair, so you'd expect that observation to have a key, value (with appropriate units etc.), and a category of "vital sign".
So in effect I'm not planning to profile out all Observations, but document them as being "types" of observation.
Yunwei Wang (Jan 21 2020 at 16:49):
Can you clarify "then have a decision tree to help decide what pattern a particular Observation conforms to." Is this decision tree for profile creator, or client/server implementer?
Dave Barnet (Jan 22 2020 at 08:42):
@Yunwei Wang Good question. Initially i'd thought it would be a profile creation decision tree (as in the UK we do a lot of push messaging rather than RESTful interactions), but I guess that once you've decided on the pattern used for an Observation you could harness ObservationDefinition to define a minimum viable construct for the Observation that could be used by a server implementor. So for the above example from Domenico the decision tree could say that a Waist Circumference is a vital sign that has a single key-value pair, and an ObservationDefinition could be created as
<ObservationDefinition xmlns="http://hl7.org/fhir">
<categoey>
<coding>
<system value="http://hl7.org/fhir/observation-category"/>
<code value="vital-signs"/>
</coding>
</category>
<code>
<coding>
<system value="http://loinc.org"/>
<code value="8280-0"/>
<display value="Waist Circumference"/>
</coding>
</code>
<identifier>
<system value="MyServer\ObservationTypes"/>
<value value="Waist Circumference"/>
</identifier>
<permittedDateType value="integer"/>
<multipleResultsAllowed value="false"/>
<quantativeDetails>
<unit>
<system value="http://unitsofmeasure.org"/>
<code value="cm"/>
<display value="centimeter"/>
</unit>
</quantativeDetails>
</ObservationDefinition>
Am I just being a bit naive here?
Kevin Mayfield (Jan 22 2020 at 10:33):
You do have a server that supports this already ..... https://dwp.hippodigital.cloud/ccri-fhir/R4/ObservationDefinition?name=heart&_count=20
Kevin Mayfield (Jan 22 2020 at 10:35):
Ooops .... This one https://data.developer.nhs.uk/ccri-fhir/R4/ObservationDefinition but it doesn't have any data in.
Grahame Grieve (Jan 22 2020 at 11:41):
there is other interest in this from implementers. I'll try to move the discussion along next week with a blog post
Dave Barnet (Jan 22 2020 at 12:09):
@Grahame Grieve That would be good. I'm at the Sydney WGM, so maybe we could have a catch up there as well
Thanks
James Walker (Jan 29 2020 at 09:10):
I have interest from the following NHS organisations Ambulance Services, Acute Hospitals, specialist Children's hospitals as well as suppliers of Obs device manufacturers, Ambulance EPR vendors and Acute hospital EPR vendors. Be good to get clarity on how to move this forward on a standards based footing
Kevin Mayfield (Jan 29 2020 at 09:51):
@James Walker what is the use case? Is this exchanging observation results between providers and/or systems.
Kevin Mayfield (Jan 29 2020 at 09:52):
i.e. an ambulance (or device in ambulance) will send results to ambulance service. Ambulance service may send some of these results to an ED.
Kevin Mayfield (Jan 29 2020 at 09:53):
an ED may wish to query observations from an ambulance service.
Kevin Mayfield (Jan 29 2020 at 09:55):
All three have different solutions, device to EPR probably a FHIR Message, ambulance service to ED probably a FHIR Document, ED looking up Observation FHIR RESTful query.
James Walker (Jan 29 2020 at 10:53):
There are a number of use cases. Device to EPR, Ambulance to A&E, Ambulance device to A&E, Aute internal transfers, Acute to Tertiary transfers
Kevin Mayfield (Jan 29 2020 at 13:35):
:+1: I think I need to look at this in more detail.
John Moehrke (Jan 29 2020 at 14:51):
There are three implementation guides that IHE has produced on this topic using FHIR. One of them is Routine Interfacility Patient Transport (RIPT) https://www.ihe.net/uploadedFiles/Documents/PCC/IHE_PCC_Suppl_RIPT.pdf, Paramedicine Care Summary https://www.ihe.net/uploadedFiles/Documents/PCC/IHE_PCC_Suppl_PCS.pdf, and Transport Record https://www.ihe.net/Technical_Framework/upload/IHE_PCC_Suppl_TransportRecordSummaryProfiles_Rev1-1_TI_2011-09-09.pdf
You can get more detail from the author Andrea afourquet@ehealthsign.com
Grahame Grieve (Feb 05 2020 at 02:50):
ok, we're going to have this discussion at 5:30pm tomorrow. I'll announce a place tomorrow.
So that is: we will be discussing ways to say which profiles apply for which subset of resources - probably (or at least mainly) in the context of CapabilityStatement
Grahame Grieve (Feb 05 2020 at 05:26):
this will be in Convention 3.3 at 5:30pm
Nick George (Feb 05 2020 at 05:29):
I take it FHIR-I in 5.10 was cancelled?
Grahame Grieve (Feb 05 2020 at 06:05):
The roundtable tomorrow night? or now?
Nick George (Feb 05 2020 at 06:10):
looked like there was one at E5.10 q4 today, but no one showed up
Lloyd McKenzie (Feb 05 2020 at 09:19):
The cancellation was communicated to HQ, but looks like it didn't make it to the app. The golden FHIR schedule is here: https://confluence.hl7.org/display/FHIRI/FHIR+WGM+Agendas+2020-01
James Walker (Apr 16 2020 at 11:15):
Hi @Grahame Grieve was there a further Observations meeting? Are there any notes?
Grahame Grieve (Apr 16 2020 at 11:17):
so I need to get to this
Last updated: Apr 12 2022 at 19:14 UTC
