Mapping LOINC Codes with Their Corresponding UMLS Codes

Description

This pretrained model maps LOINC codes to corresponding UMLS codes.

Predicted Entities

umls_code

Copy S3 URI

How to use


document_assembler = DocumentAssembler()\
    .setInputCol('text')\
    .setOutputCol('doc')

chunkAssembler = Doc2Chunk()\
    .setInputCols("doc")\
    .setOutputCol("loinc_code")

chunkerMapper = ChunkMapperModel.pretrained("loinc_umls_mapper", "en", "clinical/models")\
    .setInputCols(["loinc_code"])\
    .setOutputCol("mappings")

mapper_pipeline = Pipeline(stages=[
    document_assembler,
    chunkAssembler,
    chunkerMapper
])

data = spark.createDataFrame([["LA26702-3"],["LP99998-4"]]).toDF("text")

result = mapper_pipeline.fit(data).transform(data)


document_assembler = nlp.DocumentAssembler()\
    .setInputCol('text')\
    .setOutputCol('doc')

chunkAssembler = nlp.Doc2Chunk()\
    .setInputCols("doc")\
    .setOutputCol("loinc_code")

chunkerMapper = medical.ChunkMapperModel.pretrained("loinc_umls_mapper", "en", "clinical/models")\
    .setInputCols(["loinc_code"])\
    .setOutputCol("mappings")

mapper_pipeline = nlp.Pipeline(stages=[
    document_assembler,
    chunkAssembler,
    chunkerMapper
])

data = spark.createDataFrame([["LA26702-3"],["LP99998-4"]]).toDF("text")

result = mapper_pipeline.fit(data).transform(data)


val document_assembler = new DocumentAssembler()
      .setInputCol("text")
      .setOutputCol("document")

val chunk_assembler = new Doc2Chunk()
      .setInputCols("document")
      .setOutputCol("umls_code")

val chunkerMapper = ChunkMapperModel
      .pretrained("loinc_umls_mapper", "en", "clinical/models")
      .setInputCols(Array("umls_code"))
      .setOutputCol("mappings")

val mapper_pipeline = Pipeline().setStages(Array(
                                            document_assembler,
                                            chunk_assembler,
                                            chunkerMapper))

val data = Seq("LA26702-3","LP99998-4").toDF("text")

val result = mapper_pipeline.fit(data).transform(data)

Results


+----------+---------+
|loinc_code|umls_code|
+----------+---------+
|LA26702-3 |C0004057 |
|LP99998-4 |C0050078 |
+----------+---------+

Model Information

Model Name: loinc_umls_mapper
Compatibility: Healthcare NLP 5.5.1+
License: Licensed
Edition: Official
Input Labels: [ner_chunk]
Output Labels: [mappings]
Language: en
Size: 3.4 MB

References

Trained on concepts from LOINC for the 2024AB release of the Unified Medical Language System® (UMLS) Knowledge Sources: https://www.nlm.nih.gov/research/umls/index.html