exportMiADECDB: Exports CDB files for MedCAT / MiADE

In anoopshah/Rdiagnosislist: Manipulate SNOMED CT Diagnosis Lists

Exports CDB files for MedCAT / MiADE

Description

Produces a set of files for the findings / problems algorithm of MedCAT and MiADE. Uses the CDB environment created using createCDB which can incorporate additional manual synonyms or synonyms from WordNet.

Usage

exportMiADECDB(
  CDB,
  export_folderpath,
  lang_refset_files = NULL,
  exclude = NULL,
  include = NULL,
  exclude_historic = descendants("Disorder", SNOMED = getSNOMED()),
  blacklist = NULL,
  SNOMED = getSNOMED()
)

Arguments

CDB	concept database environment created by createCDB
export_folderpath	folder path to export to
lang_refset_files	character vector of file paths to SNOMED CT language refset files, in order to identify the preferred term for each concept. If NULL, the Fully Specified Name minus the semantic type suffix is used as the preferred term (e.g. if the Fully Specified Name is ‘Cancer (disorder)’, the default preferred term is ‘Cancer’.
exclude	a SNOMEDconcept or SNOMEDcodelist object specifying concepts to exclude from the concept database. By default, all concepts in the FINDINGS, CAUSES, BODY, LATERALITY, MORPH, SEVERITY, STAGE and QUAL tables will be included.
include	a SNOMEDconcept or SNOMEDcodelist object specifying additional concepts to include in the concept database. By default, all findings are included for potential export, but there may additional concepts of other semantic types (e.g. situation concepts) that need to be included. Inclusion takes place after exclusion, i.e. a concept in both the include and exclude lists will be included.
exclude_historic	a SNOMEDconcept or SNOMEDcodelist object specifying concepts to be excluded from the ‘historic’ lookup, i.e. those that should not be converted into historic forms. The default is to not do this conversion for disorders, only for procedures.
blacklist	a SNOMEDconcept or SNOMEDcodelist object specifying concepts to filter out of the final output. By default, concepts in the CDB of any semantic type other than ‘finding’ or ‘disorder’ are excluded. The blacklist can be used to exclude a subset of findings or disorders that are not useful for the particular application.
SNOMED	environment containing a SNOMED dictionary

Details

The following files are exported:

For MedCAT (named entity recognition and linking):

problems_cdb.csv

- CSV file in MedCAT concept database format containing cui (SNOMED CT concept ID), name, name_status ('P' for preferred term, 'N' for terms that must be disambiguated (e.g. acronyms or short terms), 'A' for synonym), ontologies = SNO (for SNOMED CT)

For MiADE postprocessing:

negated.csv

- CSV file with columns findingId (SNOMED CT concept ID of the underlying finding / disorder) and situationId (SNOMED CT concept ID of the pre-coordinated situation concept for negation of the finding / disorder). Sorted by findingId.

historic.csv

- CSV file with columns findingId (SNOMED CT concept ID of the underlying finding / disorder) and situationId (SNOMED CT concept ID of the pre-coordinated situation concept for 'history of' the finding / disorder). Sorted by findingId.

suspected.csv

- CSV file with columns findingId (SNOMED CT concept ID of the underlying finding / disorder) and situationId (SNOMED CT concept ID of the pre-coordinated situation concept for 'suspected' finding / disorder). Sorted by findingId.

overlap.csv

- CSV file with columns findingId (SNOMED CT concept ID of the underlying finding / disorder) and otherId (SNOMED CT concept ID of a concept with the same description but a different semantic type, typically a morphologic abnormality). Sorted by otherId.

problem_blacklist.csv

- CSV file without header with one column containing SNOMED CT concept IDs for concepts that may be identified by MedCAT as part of text analysis but should not be included in final MiADE output, Examples include procedure codes which may be used to link to precoordinated ‘history of...’ concepts. This file can also be used to force MiADE to ignore any specific SNOMED CT concepts in the output. Sorted in ascending order.

For more information about MiADE, visit https://www.ucl.ac.uk/health-informatics/research/miade/miade-software-and-availability

For more information about MedCAT, visit https://github.com/CogStack/MedCAT

Value

TRUE if successful

See Also

createCDB, downloadWordnet, downloadOrphanet, MANUAL_SYNONYMS, exclude_irrelevant_findings

Other MiADE functions: exclude_irrelevant_findings()

Examples

# Not run
# exportMiADECDB(CDB, export_folderpath = tempdir())

anoopshah/Rdiagnosislist documentation built on March 29, 2025, 10:46 a.m.