Stream: Orders and Observation WG
Topic: Specimen.subject - Substance question
Ralf Herzog (Jun 18 2018 at 16:25):
Hi all,
is it ok to use the Specimen.subject to model the recursive nature of containers or how this could be done?
Thanks in advance Ralf
Lloyd McKenzie (Jun 18 2018 at 17:19):
@Eric Haas? (My take is that using subject would be inappropriate, but not sure if a correct mechanism has been defined)
Eric Haas (Jun 18 2018 at 21:42):
Subject is not for the container and is inappropriate.
try
Specimen.parent Element Id Specimen.parent Definition Reference to the parent (source) specimen which is used when the specimen was either derived from or a component of another specimen. Control 0..* Type Reference(Specimen) Comments The parent specimen could be the source from which the current specimen is derived by some processing step (e.g. an aliquot or isolate or extracted nucleic acids from clinical samples) or one of many specimens that were combined to create a pooled sample.
Ralf Herzog (Jun 19 2018 at 08:26):
Your suggestion is not exactly what I am looking for. My problem relates to the need to model something like a specimen is in a container who is on a tray (with a certain position), which is on a rack (with a certain position), ... (basically v2 SAC-10ff) so how to model this recursive nature (specimen in container on a tray in a rack in ...) would this needed to be done as an extension or should it be in the resource?
Jose Costa Teixeira (Jun 19 2018 at 08:27):
Are you referring to a specific specimen, or the definition of a specimen?
Ralf Herzog (Jun 19 2018 at 08:28):
To a specific Specimen.
Jose Costa Teixeira (Jun 19 2018 at 08:29):
hmmm ok. i read this as if it were definitional, in which case specimenDefinition might have helped
Ralf Herzog (Jun 19 2018 at 08:46):
SpecimenDefinition would help, but SpecimenDefinition should be included as a possible Reference in Specimen. Since we are currently only prototyping I would like to use the SpecimenDefinition and perhaps just create an extension to Specimen to Reference it.
Does anything speaks against this (except the obvious that it is from R4 draft)?
Jose Costa Teixeira (Jun 19 2018 at 09:37):
Let's check what @François Macary has to say about this
Lloyd McKenzie (Jun 19 2018 at 10:25):
Ok, so this is about automated tracking. I don't think SpecimenDefinition would be helpful here as this is about managing a specific instance, not a 'kind'. The racks and trays would not be "specimens", so this would not be a recursive relationship. The racks and trays are (typically reuseable) devices. And device (now) does have a hierarchical relationship - and a location. If Specimen.container included a Reference(Device), would that meet your needs?
Eric Haas (Jun 19 2018 at 10:30):
...or extensions as you suggest which may be easier for and I wold certainly make a tracker for OO to look at this in FHIR because the specimen working group is certainly familiar with this. @Lorraine Constable ?
Ralf Herzog (Jun 19 2018 at 11:36):
A Reference(device) in Specimen would be a great way to do it. But it should be named appropriate. I would think about something like "carrier".
Jose Costa Teixeira (Jun 19 2018 at 11:43):
after reworking, i expect that the resource that would contain location would be the device instance, not the device kind.
Jose Costa Teixeira (Jun 19 2018 at 11:44):
but since a device instance can point to a device kind, the coverage will remain.
Jose Costa Teixeira (Jun 19 2018 at 11:46):
device (kind) will still have a hierarchical relationship (parent) to another device (kind)
Jose Costa Teixeira (Jun 19 2018 at 11:46):
and of course any of those can have an instance
Lloyd McKenzie (Jun 19 2018 at 12:51):
@Ralf Herzog, Device is used for all kinds of devices - those you're interested in but also software, complex medical devices, etc. We don't create custom-named resources for each use-case they're used in as that would result in thousands of them. If you choose to profile the Device resource for this specific use-case, you can name the profiles whatever you wish. However, in the instances, the resource would still be named "Device".
Lorraine Constable (Jun 19 2018 at 16:43):
The specimen work group is certainly familiar with this, and I understand there is a work stream focused on specimen banking use cases. Now that the domain analysis model is ready for publication, the work group intended to focus on the specimen resource(s) as the current resource is a mixture of specimen, specimen tracking and specimen collection procedures. @Rob Hausam or @Riki Merrick are most familiar with the project work and planned next steps
Peter Scholz (Nov 29 2018 at 10:45):
May I bring this topic back into use ,
in the context of biobanking for clinical trials we came across the question of container recursion too, but there also is a second relationship of containers : a specimen container may belong to a so called sample kit which means a composition of different sample containers belonging to one single collection.
So my question: Is there any ongoing work to add these relations (recursion as well as aggregation) to Specimen ?
Eric Haas (Nov 29 2018 at 15:44):
@Riki Merrick ?
Riki Merrick (Nov 29 2018 at 15:55):
Yes there is work going on - we are currently trying to define handling of derived specimen in the Tuesday specimen calls (3-4 PM ET) using the ord free-conference call line (https://join.freeconferencecall.com/ord, Online Meeting ID: ord, calls (United States): (515) 739-1430, Access Code: 294586) - we also looked at dealing with specimen event tracking (knowing where the specimen is - and are considering using the provenance resource for that. The specimen resource in FHIR definitely needs an overhaul! Please join the specimen list serve to stay in touch.
Last updated: Apr 12 2022 at 19:14 UTC
