Stream: Orders and Observation WG
Topic: Device and Calibration Data
Grahame Grieve (Jun 15 2019 at 19:03):
See https://chat.fhir.org/#narrow/stream/179166-implementers/topic/Encoding.20artificial.20pancreas.20decision.20data for an interesting question about openAPS data.
Grahame Grieve (Jun 15 2019 at 19:06):
I would like to ask a question of OO: what is the expectation for device and calibration data like some of that included in the openAPS data packet? Is it expected to be Observation.component? or some of it has a home in DeviceMetric, but it's not obvious what the link between Observation and DeviceMetric should be
@Rob Hausam @Eric Haas @Hans Buitendijk @Riki Merrick (Also @Todd Cooper @John Rhoads @Chris Courville )
Grahame Grieve (Jun 18 2019 at 20:08):
.. ping?
Rob Hausam (Jun 18 2019 at 20:44):
Still thinking. :) And actually, unless I missed something, the discussion in the linked chat didn't seem to help me much (if at all) in understanding or answering the questions.
Grahame Grieve (Jun 18 2019 at 20:46):
no let's just stick to principles here
Jose Costa Teixeira (Jun 18 2019 at 20:59):
conceptually, configuration and calibration data could be in deviceMetric, or device.property. But I do not think that those provide adequate carriers. Observation is a generic carrier for everything, so technically it works, but in this case I would avoid that.
Grahame Grieve (Jun 18 2019 at 21:01):
I think that would be called "no dollar any way" (as opposed to 'a dollar each way')
Jose Costa Teixeira (Jun 18 2019 at 21:01):
the results of calculation could be also in deviceMetric or observation. For this, I could imagine Observation as a good vehicle.
Jose Costa Teixeira (Jun 18 2019 at 21:07):
what does DeviceMetric have that Observation does not? if DeviceMetric should be used to convey heart rate, where does that leave deviceMetric?
Grahame Grieve (Jun 18 2019 at 21:07):
device metric = information about the device producing the observation
Grahame Grieve (Jun 18 2019 at 21:07):
looks like we should define an extension to link to the metrics for the device
Jose Costa Teixeira (Jun 18 2019 at 21:16):
you mean an extension on Device?
Jose Costa Teixeira (Jun 18 2019 at 21:19):
if so, then we could do that (and check overlap between that and device.property while we are at it).
Grahame Grieve (Jun 18 2019 at 21:25):
extension on Observation
Rob Hausam (Jun 18 2019 at 21:42):
Wouldn't Observation.device do much or all of what is needed? That is the link between Observation and DeviceMetric, as DeviceMetric is one of the reference choices.
Jose Costa Teixeira (Jun 18 2019 at 21:42):
I see what you mean. A direct link as an extension (but my Euro would be that the link between those is the Device: an observation is made by a device which happens to be have some configuration information given by metric)
Jose Costa Teixeira (Jun 18 2019 at 21:43):
right Rob, it could be, if we only heed the metric, and not the device itself (i don't know that use case).
Rob Hausam (Jun 18 2019 at 21:45):
Yes, as it is we have to choose between Device and DeviceMetric, as it's 0..1 - but DeviceMetric also links to the source and parent Device.
Grahame Grieve (Jun 18 2019 at 21:48):
well, ok, I had not noticed that
Jose Costa Teixeira (Jun 18 2019 at 21:51):
ok then i probably misread the DeviceMetric intent.
Eric Haas (Jun 18 2019 at 21:51):
DeviceMetric's definition is : "Describes a measurement, calculation or setting capability of a medical device." So It sounds to me like all that stuff should live in there ( whether is actually is capable of representing it right now notwithstanding) So Observation.device --> DeviceMetric --> Device. Looks like how it was intended to be modelled but I have nto been active with devices for a while so maybe that thinking has changed.
Eric Haas (Jun 18 2019 at 21:51):
can choose the right color wire with DeviceMetric :-)
Jose Costa Teixeira (Jun 18 2019 at 21:52):
Seems clear. Is DeviceMetric capable of exposing all that is needed for the artificial pancreas?
Eric Haas (Jun 18 2019 at 21:54):
I don't see that Jose need to get the Device folks on the horn and explain it. I would assume need a name value pair somewhere to describe the parameters??
Jose Costa Teixeira (Jun 18 2019 at 21:58):
indeed. To me, Device Metric reads (now) as "the current configuration for a device" but it does not contain a placeholder for things like "activation delay, BG target range, BG low threshold"...
Jose Costa Teixeira (Jun 18 2019 at 22:00):
We must catch up on Device, so we should bring this requirement, and see where it fits - DeviceMetric, or device.property of we keep that one...
Andrea Pitkus, PhD, MLS(ASCP)CM, CSM (Apr 29 2020 at 13:13):
Would the question also apply for laboratory QC and Calibration data? Calibration is more to instrument, but QC is to instrument run during certain time period as you know. For some tests like POCT, it's entered as an observation, sometimes to note that QC passed, say for PT/INT testing.
Last updated: Apr 12 2022 at 19:14 UTC
