OHDSI Vocabulary / Population Browser

  1. overall goals:
    1. make individual OHDSI vocabularies more approachable
    2. facilitate understanding of OHDSI's vocabularies and concepts, relationships between them, and their prevelance in a specific CDM
    3. facilitate translation from individual source concepts for cohort or phenotype definition and for ETL work
    4. facilitate development or translation of concept sets
    5. provide tools and interface for classification/granularity selection in other contexts (e.g., patient profiles)
    6. towards these ends:
      1. develop detailed requirements, designs, and prototypes
      2. implement in ATLAS if and when resources are available
  2. use cases:
    1. define a cohort or phenotype (needed by ATLAS users):
      1. translate existing phenotype definition (e.g., set of ICD-9 codes for some condition) from a source vocabulary into a target vocabulary
    2. explore the whole set of OHDSI vocabularies:
      1. needed in Athena
      2. to improve and debug the vocabularies
      3. in general
      4. only for concepts in user's CDM
      5. for a specific domain:
        1. e.g., provider specialty, show record counts for all values including zero counts for specialties not used; maybe also show corresponding source values and source values that weren't able to be mapped to a target. (provider.specialty_concept_id, provider.specialty_source_value, provider.specialty_source_concept_id, care_site.specialty_concept_id, care_site.specialty_source_value)
      6. to better understand relationships between source and target domains
      7. to better understand classifications available
    3. source(s) to target(s):
      1. find mapped or matching codes, classification or standard
      2. in target code(s) selected, see which source codes are added or left out and counts of corresponding patient records
    4. find a concept:
      1. have immediate feedback on CDM records with concept
      2. see relationships to other concepts and be encouraged to assure one has chosen correctly
    5. explore my popupaltion:
      1. overall
      2. in terms of some domain
      3. defining a subpopulation (same as 1)
      4. intersecion of subpopulations
    6. Patrick's SNOMED navigation strategy:
      1. using related concepts filter on Class, choose Has ancestor of from Relationship facet, filter on Distance facet
    7. compare vocabulary version as per Peter's post
  3. notes and to-dos:
    1. other systems to get ideas from:
      1. http://outins.com/
      2. http://jigsaw.io/
      3. http://www.ephir.com/
      4. http://www.aviz.fr/Research/Nodetrix
    2. javascript libraries to consider using:
      1. https://www.golden-layout.com/examples/
    3. to-dos:
      1. integrate ERD maybe:
        1. http://www.ohdsi.org/web/wiki/lib/exe/fetch.php?media=documentation:omop_cdm_v5_erd.pdf
      2. source record count:
        1. show these as existing in cdm but not available through query
        2. also show ancestor concepts with their descendant record counts
        3. only show classification concepts that have a descendant record count
      3. default to showing 4 main domains (condition, drug, procedure, measurement)