Closed comorbidity closed 5 months ago
OK, maybe we could call this a "label group" and have a config like this:
label-groups:
positive-case: [ideation-past, ideation-present, action-past, action-present]
And that group could show up in the math... But how? As just an extra presentation-only row in the output (with no effect on F1), or should it (sometimes?) replace the individual labels in the matrix and be used in the F1 calculation?
Will be solved by #18
similar to "label implies", the "binary label" would collapse all labels into one.
For example, if we wanted to know all the patient notes that have any label of "suicidality"
ideation-past ideation-present action-past action-present
would all map to the label "positive-case" (here CASE means case definition, not "case" as an IF/Then/Else)