OHDSI Home | Forums | Wiki | Github

The same measurement from a different device

Hello, I am Jin Choi from Ajou University.
I am working on research linking bio-signals and CDM.

Situation :
Recently we want to measure the QT interval from electrocardiogram data from a continuous patient monitoring device(ICU).

problem :
The QT interval was traditionally measured only on a 12-electrode ECG.
Accordingly, existing QT interval concepts seem to assume a 12-electrode electrocardiogram (not including data about measurement method).
In the future, More diverse devices will generate ECG data (patient monitoring devices, wearable devicesā€¦etc.)

Question:
What should I do if these same measurements are from different devices?

Possible solution(In my opinion):

  1. Among existing concepts, find and use the most similar ones
  2. Create a new extended term (like ā€˜QT interval from patient monitoring deviceā€™)
  3. Insert information about measurement method into Measurement_source_concept_id
  4. Create a separate column for the measurement method in the measurement table

Iā€™m going to choose solution 1 right now. Accordingly, the QT interval from the patient monitoring device will be mapped to 3004627 (QT interval Lead II, LOINC)

In my opinion, number 2 seems to be the most appropriate solution, but the process can be slow. In the future, if various devices such as wearable devices are included in the CDM, there may be a problem of generating a huge number of terms.

Number 3 is an easy solution, but it seems to go against the schema of the CDM.

Number 4 seems most ideal to me, but there seems to be a problem with changing the CDM schema.

Is there any discussion about this problem?

I think people working with MIMIC WFDB will have similar problems.
@mkwong

You might want to add a record in the Device table and then link it to the Measurement record to distinguish the Devices

If you are using a patient monitor, wearable, or other non-standard 12-lead ECG device to generate QT, I would suggest considering using concept LOINC.8634-8 (OMOP Concept.concept_id = 3025809 ā€œQ-T intervalā€) if the lead is not one of the standard 12-lead leads (I-III, aVR, aVL, aVF, V1-V6).

While this may stretch the technical use definition - I often then include the device identifier in the measurements.measurement_source_value field to link it back to the originating device if multiple devices can produce a QT measurement for the same patient/visit at the exact same date/time.

The above LOINC code is what MIMIC-IV use to map the QT trend/time-series data if the lead label does not fall into the standard 12-lead label set.

Agree with @MPhilofsky . You can choose the general device-independent concept for your measurement and store the information about a specific device in the device table. Then you can link those two events somehow, like using the fact_relationship table or the Oncology extension approach. The latter requires the measurement table to be extended with two additional fields - modifier_of_event_id and modifier_of_field_concept_id. Since you are using the Device table for storing the device information, the modifier_of_field_concept_id will be 1147115 (in the case of OMOP CDM v5.3.1), which is device_exposure.device_exposure_id. The actual device_exposure_ids are stored in the modifier_of_event_id correspondingly. See https://ohdsi.github.io/CommonDataModel/oncology.html#MEASUREMENT

1 Like

Another option we have kicked around regarding linking the device_exposure and measurement is through the use of device_exposure.visit_detail_id = measurement.visit_detail_id. It is an attractive approach but no consensus yet.

Thank you for your advice.

In fact, we mapped the QT value from the standard 12-lead ECG to 3025809.

How do you map the QT value from the standard 12-lead ECG?

@MPhilofsky @Philip_Solovyev

Thanks for your advice.

As you advised, it would be appropriate to use the fact_relationship table to represent the relationship between the measurement and the device.

1 Like

For standard 12-lead QT measurements - query the pattern ā€œQ-T interval | Lead V2 | EKG measurementsā€ in the concept table. Replace V2 with other standard 12-lead ECG labels.

t