To get started building your cohort, start by selecting a filter (aka attribute) made available by your selected data source. A filter can be added to the “Match Any” section, the “Match All section, or to both.
Vertical vs Horizontal Cohorts
If you consider the result of a cohort to be a data frame, a horizontal cohort is one that relates to the columns of a data frame. Columns run horizontally across the data frame and represent unique data modalities. A vertical cohort is one that relates to the rows of the data frame. Rows run vertically across the data frame and represent unique patients. The patients in a data frame can be unique per data source or globally unique across the federation.
A horizontal cohort cannot be constructed without first building a vertical (seed) cohort. A seed cohort is a set of patients that the user is interested in expanding their data coverage for. Thus, the result of the horizontal cohort is the subset of patients from the seed cohort that return true for the data modalities requested in the horizontal cohort definition.
Add a Horizontal Group
After the initial vertical (seed) group has been created, click the "Add Horizontal Group" button. When adding filters to this horizontal group, a venn diagram will display, illustrating the total patient overlap between the two groups.
Add an Additional Vertical Group
Group A displays by default, but one additional vertical group can be added when creating a cohort. This allows you to write intersect, subtract, and union queries.
Lock Cohort
Since the data sets are “online” and dynamic, your cohort may return slightly different results every time the Cohort page is loaded. If you no longer want your data set to behave dynamically for a specific Cohort Query, you can lock the query to save the data in its current state.
Note that the cohort query will no longer be editable after it is locked and that this action can not be undone. After a cohort query is locked, it can be copied if a dynamic data set is desired.
