OHDSI Home | Forums | Wiki | Github

Vocabularies for wearables and telemetry data

CDM Builders
,
#1

As a pet project, I’m researching the best practices for developing standard vocabularies for wearables and other telemetry data. I know that the [Pistoia Alliance is pushing for more collaboration to develop device standards][1]. I’m not sure if there are folks in this community directly plugged into this (@keesvanbochove).

Has anyone built their own CDM extensions to house these kind of data? (@Patrick_Ryan @Frank @jon_duke @Christian_Reich @Rijnbeek @nigam ) It know it’s the wild west in terms of standardization. Looking for any and all war stories to collate and build a robust point of view. I also remember seeing @JackPo mention some work linking these data to other EHR data but haven’t seen other recent discussions.

If anyone has any papers they’ve read, send them my way. I’m just in an information gathering stage at present. My ambitious, longer term goal would be to try to create a strong enough point of view to inform a proposal to the CDM working group but I’m nowhere near that point. :smiley: [1]: http://www.clinicalinformaticsnews.com/2017/05/03/come-together-the-pistoia-alliances-push-for-wearable-device-standards-in-clinical-trials.aspx

2 Likes
Remote (telemetry) BP readings
#2

Hi Kristin,

That’s a cool question, right at the intersection of two of our hottest tools, OHDSI and RADAR :smile:
There are some standards for this (there’s IEEE-11073 but that’s very hardcore device centric, there’s a section in the Bluetooth protocol, there’s some FHIR profiles building - Observation - FHIR v4.6.0) etc. and of course lots of startups building solution for this like HumanAPI. In RADAR-CNS we’re working on a fully open source stack and we have a MongoDB and API for querying the data in a warehouse like fashion, and the raw data is in HDFS. For sharing the data we for now use Avro files, which have the schema embedded, but are also looking at more options.

It would seem to me that in OMOP, the Measurement domain would be the appropriate place to put it, not much modifications needed to CDM except for some database tuning (e.g. partitioning) once you start putting a lot of raw data.

But the question if it makes sense to put these two types of data together. Depending on the question you would like to ask the integrated medical history and sensors data, it would probably make more sense to store some kind of aggregation or ‘era’ in OHDSI terms, in which case a CDM extension might make sense.

Greetings,

Kees

#3

I remember seeing a poster about wearable data at one of the OHDSI symposia?!

There is the RADAR project in Europe (https://www.imi.europa.eu/content/radar-cns) that @Kees Van bochovenmailto:kees@thehyve.nl is involved in and OMOP-CDM is most probably on the RADAR :relaxed:?

Peter

#4

Thanks @keesvanbochove and @Rijnbeek Thanks for putting RADAR on my radar. :laughing:

#6

We are now collecting a lot of realtime biosignal data form 80 ICU monitoring devices at Ajou Unviersity Hospital in South Korea. We are now try to integrate them into CDM, however also at a very early stage in the consideration. @yoon8302 is in charge of it.

1 Like
#8

Hello. I find this thread very interesting. I also read about RADAR and other projects.

@krfeeney: I’m wondering, are you closer to your longer-term goal or have you found another way to better incorporate that kind of data into the CDM?

Mathias

t