Source code/value explorer Dashboard

anthonysena / DrugExposureCharacterization

Apache License 2.0

0 stars 0 forks source link

Source code/value explorer Dashboard #6

Open PRijnbeek opened 4 years ago

PRijnbeek commented 4 years ago

Would be good if we can add a dashboard that explores the source codes and values and the mapped to standard concepts.

This would provide a good quality control mechanism.

anthonysena commented 4 years ago

I've added a new dashboard called "Source Code Explorer" which I think should be a good place to start this one.

cgreich commented 4 years ago

Definitely. We need to think about how to capture a few issues:

Outdated vocabulary files (mappings available centrally vs locally)
Local source codes adn their descriptions

anthonysena commented 4 years ago

Outdated vocabulary files (mappings available centrally vs locally)

The export function of the package will query the CDM's vocabulary table and include the vocabulary version as part of the metadata for the source. Any other thoughts on what else may help us indicate outdated vocabulary files?

Local source codes and their descriptions

Some ideas here:

Look at exporting the source_to_concept_map table entries that match on the concept_id's that are used in the drug exposure summary
Use the concept_synonym table in the vocab to identify potential concepts that could be considered from the local source codes given a language_concept_id. Admittedly, I haven't use this table int he course of my work...

cgreich commented 4 years ago

Any other thoughts on what else may help us indicate outdated vocabulary files?

You could download the source_concept_id to concept_id maps from the source, and then compare to see if something changed.

source_to_concept_map table entries that match on the concept_ids

Exactly.

potential concepts that could be considered from the local source codes

I wouldn't go there. This is about reporting, debugging and QA. Not about fixing.

anthonysena commented 4 years ago

You could download the source_concept_id to concept_id maps from the source, and then compare to see if something changed.

Good news is that we're already doing this - we capture this drug_source_concept_id <-> drug_concept_id mapping in the "dus_de_detail" table.

Look at exporting the source_to_concept_map table entries that match on the concept_id's that are used in the drug exposure summary

One more question here: any concerns by sites if this type of information is extracted? Generally this is something we use internally for ETL and keep it around for QC.