OHDSI Home | Forums | Wiki | Github

Aggregate data extraction?

Developers
#1

Greetings,

Are there means within the OHDSI platform to extract aggregate data?

A use case to help explain: Density of diabetes diagnoses for each census tract in Maine (assuming the mapping from location to tract already exists). The end goal is to allow the user to extract aggregate patient data into geographic tables (county, tract, etc) to be used for maps and other visualizations.

As I understand it, the typical OHDSI process would be to define a cohort and then do your feature extraction. This doesn’t seem applicable here as creating a distinct cohort for each tract would produce hundreds of cohorts.

If I am correct in my assumption that new tools will need to be developed to perform this extraction, are there specific guidelines in place regarding how a user should interact with concepts? For example, should concept sets be incorporated here as well?

I understand that this a loaded and ambiguous question but I figured I would throw it out there before we dive too deep into it.

#2

@rtmill:

There are probably many ways to Rome, but why don’t you create one cohort with diabetes patients in Maine, and then use their detailed geographical information to stratify the members of the cohort by that geography and count them up? If you need Concept Sets in that definition the current tools support that no problem.

#3

I used density as an example to try and show how this can be tricky. In that case, we would need counts of both diabetes patients as well as total patients for each region. We could create a diabetes cohort and compare the results to census data or something similar but I don’t believe it would be as descriptive.

t