APPLICATION

Mapping is a tool for mapping identifiers (ID, codes) between clinical ontologies and lexical resources.

The tool interlinks identifiers from UMLS, SMOMED-CT, ICD-10, the corresponding Wikipedia articles and WordNet synsets. It's main goal is to provide semantic interoperability across the clinical concepts from various knowledge bases.

Requirements and Installation

1. Before starting, the database should be indexed in Elasticsearch. The database is under UMLS and SNOMED CT licences. https://www.nlm.nih.gov/research/umls/knowledge_sources/metathesaurus/release/license_agreement_snomed.html

URL

2. The database will be uploaded and indexed in Elasticseach, which should be either installed on your device or used as a Docker image. Here, we install Elasticsearch as a Docker application with the following command.

docker-compose up

Stop and delete Docker application:

docker-compose down

uvicorn application.web.main:create_app --host 0.0.0.0 --port 5858 --reload

API use

The API has three methods

1. application/{index_name} Post Index - method for database indexing

Input format contains the source ID we want to map, the type of the taxonomy and the flag if we need to get the infromation about this ID from Wikipedia and WordNet.

The source type must me UMLS, SNOMED_CT, ICD10CM or ICD10PCS.

index_name: umls 
{
  "source_id": "1003369001",
  "source_type": "SNOMED_CT",
  "language": "SPA",
  "wiki": false
}

When the Elasticsearch API is up, we should update databases in Elasticsearch index

We pass the text file to the API. To process them correctly, the following arguments should be provided.

2. idmap/{index_name} Delete Index - method for index deleting

{
  "index_name": "string" # the name of the index where the database is indexed
  "path": "string",
  "headers": ["string"], # optional
  "separator": "string"
}

3. idmap/map Get Item Mapping - the main method for code mapping

{
  "index_name": "string"
}

CLI

To update the Wikidata codes

python -m application.wiki_wordnet.update_wiki

TODO

CLI for batch processing

Elastic commands

curl 'localhost:9200/_cat/indices?v'

curl -XDELETE 'http://localhost:9500/umls'

Examples

% UMLS C0011860 - diabetes tipo 2 C0025519 - metabolsm

% SNOMED_CT 19997007 - hipnosis

% ICD10CM H35.35 - Cystoid macular degeneration