Description
This model maps drug entities extracted from clinical text to their corresponding ATC (Anatomic Therapeutic Chemical) codes. It uses ner_posology for drug entity recognition and provides fast code mapping without requiring embeddings at inference time. Trained on the WHO ATC-DDD dataset (release 2026-04-25).
Live Demo Open in Colab Copy S3 URI
How to use
document_assembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = SentenceDetectorDLModel.pretrained("sentence_detector_dl_healthcare", "en", "clinical/models")\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
word_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")\
.setInputCols(["sentence", "token"])\
.setOutputCol("word_embeddings")
ner_posology = MedicalNerModel.pretrained("ner_posology", "en", "clinical/models")\
.setInputCols(["sentence", "token", "word_embeddings"])\
.setOutputCol("ner")
ner_converter = NerConverterInternal()\
.setInputCols(["sentence", "token", "ner"])\
.setOutputCol("ner_chunk")\
.setWhiteList(["DRUG"])
atc_mapper = ChunkMapperModel.pretrained("atc_mapper", "en", "clinical/models")\
.setInputCols(["ner_chunk"])\
.setOutputCol("mappings")\
.setRels(["atc_code"])
pipeline = Pipeline(stages=[
document_assembler, sentence_detector, tokenizer, word_embeddings,
ner_posology, ner_converter, atc_mapper
])
data = spark.createDataFrame([["The patient was started on metformin 500 mg twice daily for type 2 diabetes and was also prescribed atorvastatin for hyperlipidemia. She was given amoxicillin for a sinus infection."]]).toDF("text")
result = pipeline.fit(data).transform(data)
document_assembler = nlp.DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = nlp.SentenceDetectorDLModel.pretrained("sentence_detector_dl_healthcare", "en", "clinical/models")\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = nlp.Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
word_embeddings = nlp.WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")\
.setInputCols(["sentence", "token"])\
.setOutputCol("word_embeddings")
ner_posology = medical.NerModel.pretrained("ner_posology", "en", "clinical/models")\
.setInputCols(["sentence", "token", "word_embeddings"])\
.setOutputCol("ner")
ner_converter = medical.NerConverterInternal()\
.setInputCols(["sentence", "token", "ner"])\
.setOutputCol("ner_chunk")\
.setWhiteList(["DRUG"])
atc_mapper = medical.ChunkMapperModel.pretrained("atc_mapper", "en", "clinical/models")\
.setInputCols(["ner_chunk"])\
.setOutputCol("mappings")\
.setRels(["atc_code"])
pipeline = nlp.Pipeline(stages=[
document_assembler, sentence_detector, tokenizer, word_embeddings,
ner_posology, ner_converter, atc_mapper
])
data = spark.createDataFrame([["The patient was started on metformin 500 mg twice daily for type 2 diabetes and was also prescribed atorvastatin for hyperlipidemia. She was given amoxicillin for a sinus infection."]]).toDF("text")
result = pipeline.fit(data).transform(data)
val documentAssembler = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val sentenceDetector = SentenceDetectorDLModel.pretrained("sentence_detector_dl_healthcare", "en", "clinical/models")
.setInputCols(Array("document"))
.setOutputCol("sentence")
val tokenizer = new Tokenizer()
.setInputCols("sentence")
.setOutputCol("token")
val wordEmbeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")
.setInputCols(Array("sentence", "token"))
.setOutputCol("word_embeddings")
val nerPosology = MedicalNerModel.pretrained("ner_posology", "en", "clinical/models")
.setInputCols(Array("sentence", "token", "word_embeddings"))
.setOutputCol("ner")
val nerConverter = new NerConverterInternal()
.setInputCols(Array("sentence", "token", "ner"))
.setOutputCol("ner_chunk")
.setWhiteList(Array("DRUG"))
val atcMapper = ChunkMapperModel.pretrained("atc_mapper", "en", "clinical/models")
.setInputCols(Array("ner_chunk"))
.setOutputCol("mappings")
.setRels(Array("atc_code"))
val pipeline = new Pipeline().setStages(Array(
documentAssembler, sentenceDetector, tokenizer, wordEmbeddings,
nerPosology, nerConverter, atcMapper
))
val data = Seq("The patient was started on metformin 500 mg twice daily for type 2 diabetes and was also prescribed atorvastatin for hyperlipidemia. She was given amoxicillin for a sinus infection.").toDF("text")
val result = pipeline.fit(data).transform(data)
Results
| ner_chunk | atc_code |
|:-------------|:-----------|
| metformin | A10BA02 |
| atorvastatin | C10AA05 |
| amoxicillin | J01CA04 |
Model Information
| Model Name: | atc_mapper |
| Compatibility: | Healthcare NLP 6.4.1+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [ner_chunk] |
| Output Labels: | [mappings] |
| Language: | en |
| Size: | 811.0 KB |