OHDSI Home | Forums | Wiki | Github

OHDSI Informatics Study: Data Quality


(Vojtech Huser) #1

Dear collaborators,

I would like to formally invite any interested site to join a Data Quality study. The study was piloted at several sites (in the last few months) and we would like to make a more formal announcement and invite yet more sites to participate.

The study details can be found here:
http://www.ohdsi.org/web/wiki/doku.php?id=research:dqstudy

A link to the protocol for the study is here:

We currently have data from 10 CDM datasets (with many of the major OHDSI adopters on board), however since we are establishing empirical thresholds for some data quality indicators, we want to include as many sites as possible.

Some related threads are:

I would be happy to answer any questions (via forum or direct email).


(Lisa Schilling) #2

@Vojtech_Huser I’d like to participate!


(Vojtech Huser) #3

I am pleased to update that some of the outputs from the Data Quality study have been incorporated into a version 1.4 of achilles (updated in GitHub 4 days ago).

Some of the new version 1.4 features are:

  • new measures 10,11,12
  • new measures in 2100 range (devices) and 2200 range (notes)
  • one new DQ rule
  • classifications of rules into categories (CDM conformance rules vs. DataQuality rules)
  • updated readme file (and notes.md file in extras folder)
  • other smaller changes

Some of those new additions impact our wishlist for the web UI.
AchillesWeb is now officially referred to as Atlas-DataSources. See the issues here: https://github.com/OHDSI/Atlas/issues?q=is%3Aissue+is%3Aopen+label%3A"data+sources"


(Vojtech Huser) #4

I would like to preliminary announce that together with Yohan Sumathipala, we have created an add-on package to the DataQuality study. It is still undergoing revisions and fine tuning.

The idea is to analyze trends and unexpected trends may indicate a data quality problem with the data. The trend analysis is a sub-component of the Data Quality study (and uses the same protocol)

We are still fine tuning the level of data exported for central analysis (making the extract as minimal as possible). The package currently outputs data in two folders: (1) result folder (not to be shared, for local use by local researchers) and in (2) export folder (for central processing).

The package README is here: https://github.com/OHDSI/StudyProtocolSandbox/tree/master/OHDSITrends


(Vojtech Huser) #5

In preparation to DQ Tutorial and as part of write up of ThemisMeasurment database article, the DataQuality package was updated to include a new function that (“in dev mode”) checks units in your lab data using ThemisMeasurements database.


t