--* exclude_concepts:
--*   attr:
--*     fillcolor: '2'
--*   desc: Defines the set of concept_ids to exclude from the data; a few specific ones,
--*     and anything not in the condition domain.
--*   ext: sql
--*   inputs:
--*   - concept_lda_pin
--*   - concept_set_members_lda_pin
--* 

-- here we define concepts to filter out
-- variations of Findings of Sexual Activity (used by one only data partner that still passes QC)
-- COVID-19 itself (this has a large effect, when present it acts like an attractor, collapsing various COVID-related subtopic)
-- boring concepts (clinical finding etc.)
-- anything in the OMOP Concept table that isnt in the "condition" domain 
--  (some OMOP sites apparently add non-condition concepts into here erroneously)

    (
    SELECT concept_id, concept_name
    FROM concept_set_members_lda_pin
    WHERE codeset_id = 192528259 -- Findings of Sexual Activity (one of these is an Observation, so it will be duplicated with hte third group below)
    )
UNION ALL
    (
    SELECT concept_id, concept_name
    FROM concept_lda_pin
    WHERE concept_id = 441840   -- clinical finding
    OR concept_id = 0        -- no matching Concept
    OR concept_id = 37311061 -- COVID-19
    OR concept_id = 4086701  -- no abnormality detected
    OR concept_id = 440029 -- viral disease
    OR concept_id = 4100065 -- disease due to coronaviridae
    )
UNION ALL
    (
    SELECT concept_id, concept_name
    FROM concept_lda_pin
    WHERE domain_id != 'Condition'
    )