Hello fellow ontologists,
Burak Serdar joins us from Cloud Privacy Labs, which, with their data
warehouse partner, received a 2021 ONC LEAP award for semantic
harmonization of health data. Welcome, Burak! :)
Burak's offered to present during our next Oct. 15th meeting on the topic
of layered schemas.
"The idea is to use schemas with semantic layers that add annotations to
> data. When I
> first started I had JSONLD/RDF in mind, but I slowly moved towards a more
> general labeled property graph approach. We are using this approach to
> ingest JSON, CSV, and XML documents to convert them to OMOP through a
> semantic pipeline for our ONC LEAP project. This approach can combine
> multiple vocabularies at the schema level. Since it is a layered approach,
> different layers can be used to add terms from vocabularies to the
> underlying data based on context.
Burak further explained:
Layered schemas architecture is structurally similar to the OCA, but there
> are some fundamental differences between the two approaches. Layered
> schemas work with functional terms of a vocabulary, rather than functional
> overlays. Because of this, it is possible to merge multiple vocabularies to
> process documents. For instance, you can easily add
> w3c's data privacy vocabulary onto the FHIR schema. ...
There is a web site for layered schemas: https://layeredschemas.org
There is an article about how to add schema layers for data privacy
> vocabulary: http://layeredschemas.org/docs/dpvls/
There is a "layered schema playground" that also includes layered schema
> representation of the FHIR schema. There are other sample schemas there as
> well.
We are working on the concept of "semantic pipelines" using layered schemas
> to annotate and transform data. We have some proof-of-concept
> implementations that generate credentials from FHIR input using layered
> schemas. I am also working on an implementation of FHIR profiles using
> layered schemas.
This is the architecture we are using for our ONC project to harmonize
> healthcare data coming from multiple sources. This project is on
> harmonizing data with different representations (CSV, FHIR (JSON), XML) and
> different vocabularies.
Looking forward to learning more on Oct 15th!
Eric Jahn
CTO/Data Architect
Alexandria Consulting LLC
St. Petersburg, Florida
727.537.9474
alexandriaconsulting.com