search

OHDSI / Achilles

Automated Characterization of Health Information at Large-scale Longitudinal Evidence Systems (ACHILLES) - descriptive statistics about a OMOP CDM database
https://ohdsi.github.io/Achilles/
130 stars 122 forks source link

Heel (DQ): Percentage of non-numerical measurement records #105

Closed ericaVoss closed 8 years ago

ericaVoss commented 8 years ago

@vojtechhuser,

Hey, just ran the ACHILLES and got this error:

WARNING: percentage of non-numerical measurement records exceeds general population threshold

I see you hard coded to 80%. Where did that come from? When you say "general population threshold" it sounds like you compared against other DBs.

Either way, it may be nicer to say "WARNING: percentage of non-numerical measurement records exceeds 80%"

ericaVoss commented 8 years ago

Also, for DBs that only have procedure claims records that are considered measurements (e.g. CPT4 '85025-Blood count; complete (CBC), automated (Hgb, Hct, RBC, WBC and platelet count) and automated differential WBC count`), this error will fire. May it make sense to do something that includes looking at the type of record that is coming across?

vojtechhuser commented 8 years ago

In the StudySandbox, there is a data quality study that will collect from multiple sites this number and threshold will be decided based upon this analysis.

I am hoping J&J will participate.

The actual number is in the new table achilles_results_derived. Once you re-run achilles on datasets, the data for such DQ study will be available. I plan to add many other similar measures for which we will deside some general population range. I am also suggesting some dataset metadata that designate a datset as general population.

On Tue, May 10, 2016 at 5:41 PM, ericaVoss notifications@github.com wrote:

@vojtechhuser https://github.com/vojtechhuser,

Hey, just ran the ACHILLES and got this error:

WARNING: percentage of non-numerical measurement records exceeds general population threshold

I see you hard coded to 80%. Where did that come from? When you say "general population threshold" it sounds like you compared against other DBs.

Either way, it may be nicer to say "WARNING: percentage of non-numerical measurement records exceeds 80%"

— You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub https://github.com/OHDSI/Achilles/issues/105

vojtechhuser commented 8 years ago

Patrick suggested to have new type of output (NOTIFICATION) . So this was changed to notification.