Increase Findability of CDEs

[jkyu]

Task description:

Incorporate public CDEs from the NIH and other CDE repositories into the metadata specification tool. Enhance search functionalities to ensure the straightforward identification and retrieval of CDEs. Ensure the code is hosted in an accessible repository, accompanied by comprehensive documentation, and fully integrated into the tool. If time allows, enhance the Data Hub search engine to support finding CDEs

[jkyu]

The first item we will address is the formal specification of the CDEs used in RADx projects. We have verified that the Tier 1 CDEs used in RADx do not match any NIH-endorsed CDEs present in the NIH CDE Repository. RADx data elements are currently defined only in data dictionaries and spreadsheets, so they are are difficult to find, maintain, and reuse.

To address this, we will develop a software tool to ingest CDEs used by RADx projects and provide a CDE specification in the CEDAR JSON-LD format. We will need to ingest the contents of the global codebook for Tier-1 data elements as well as data dictionaries for Tier-2 data elements. These CDE definitions will then be stored on CEDAR, which allows us to establish a formal source of truth for CDEs with version control. These CDEs can be uploaded to the NIH CDE Repoistory at a later date.

Ingestion of the Tier-1 data elements is straightforward. Tier-2 data elements are trickier, since they are converted from data dictionaries. Data dictionaries that follow the RADx specification can be ingested with processing by the components of the data dictionary validator. Some data dictionaries do not follow the specification and require bespoke code.

[jkyu]

Developed a small package to convert and ingest CDEs into CEDAR from the global codebook (code here).

This task also includes the work to ingest CDEs from the NIH CDE Repository (code here).

Working document detailing the technical work for this task is here.