I’m looking for a way, using Loinc, to associate an observation (e.g. patient’s temperature) with the body position of the patient at the moment the observation is made.
Both measures will be made by automatic sensors.
I’m not sure what of these approaches is the more correct (or if there is a better one):
1- Create one observation for the temperature and another observation for the position.
2- Create one observation with the patient’s temperature, create a DocumentReference with the sensors’ raw output as an attachment, and then create an aggregator observation that contains the patient’s temperature and the calculated position and references the first observation and DocumentReference as derivedFrom.
I’m not sure if the Position of the patient should be considered one observation.
There are two common ways to capture body position:
- Pre-coordinate it with the code (e.g. there are distinct LOINC codes for blood pressure sitting vs. standing vs. lying)
- Capture it as a component of the base measurement - as it is a ‘qualifier’ of the Observation being made and is intrinsically tied to the measured value.
Thanks for the reply!
Since some of the sensors measures are not usually associated with the body position, there is not a Loinc code that represents this info (e.g ECG, Body Temperature, Respiration…).
For the second option, since the IMU sensor is another device and component should only be used when there is only one method, one observation, one performer, one device, and one time, I’m not sure if it fits as a solution for this problem.
Would it be correct to have an observation with the Loinc Code for body position, another with the other sensors output (e.g patient’s temperature), and another observation that refers to these two as a hasMember?
Is this a situation where you’ve just got a bunch of sensors making periodic measurements and you’re correlating them together based on time, or is there a ‘controller’ type relationship where something triggers all of the devices to capture measurements at a single time-point?
Thanks for the reply!
This project consists of having a bunch of sensors attached to the pacient that are connected to a Smartbox near the patient that is then connected to a central system.
The Central system is responsible for requesting a sensor output to a smartbox.
The SmartBox is responsible for receiving the request, detect what sensors should be triggered, making a request to these sensors, receive the output, and then send it to the central system.
These requests can ask for info on all the sensors or to just one in specific, and can be triggered by a manual user action or a timed ServiceRequest. Every time a request is made, the IMU sensor should also be triggered and return the body position in a way that is possible to associate with the values from the requested sensor.
On that basis, if there’s a need to group different sensor readings that were all triggered by one request, I’d recommend having a ‘grouper’ observation that points to the individual observation using ‘Observation.member’. Each Observation would be a separate instance. If you want to transmit them all as a single call, you can use transaction.
Thanks for your help!
In case I want to save the IMU sensor raw output as an FHIR Resource, is it correct to create a DocumentReference containing the output as an attachment and then refer to this DocumentReference from the Body Position Observation as a derivedFrom?
The output would be a JSON with the following structure:
"pitch" : 0,
If you need the raw data and converting it to Observations isn’t acceptable, then yes, you could wrap with a DocumentReference