Detect Chemical Compounds and Genes

Description

This is a pre-trained model that can be used to automatically detect all chemical compounds and gene mentions from medical texts.

Predicted Entities

CHEMICAL, GENE-Y, GENE-N

Live Demo Open in Colab Copy S3 URI

How to use

Use as part of an nlp pipeline with the following stages: DocumentAssembler, SentenceDetector, Tokenizer, WordEmbeddingsModel, NerDLModel. Add the NerConverter to the end of the pipeline to convert entity tokens into full entity chunks.

...
word_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")\
.setInputCols(["sentence", "token"])\
.setOutputCol("embeddings")
clinical_ner = NerDLModel.pretrained("ner_chemprot_clinical", "en", "clinical/models") \
.setInputCols(["sentence", "token", "embeddings"]) \
.setOutputCol("ner")
...
nlp_pipeline = Pipeline(stages=[document_assembler, sentence_detector, tokenizer, word_embeddings, clinical_ner, ner_converter])
light_pipeline = LightPipeline(nlp_pipeline.fit(spark.createDataFrame([['']]).toDF("text")))
results = light_pipeline.fullAnnotate("Keratinocyte growth factor and acidic fibroblast growth factor are mitogens for primary cultures of mammary epithelium.")

...
val word_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")
.setInputCols(Array("sentence", "token"))
.setOutputCol("embeddings")
val ner = NerDLModel.pretrained("ner_chemprot_clinical", "en", "clinical/models")
.setInputCols("sentence", "token", "embeddings") 
.setOutputCol("ner")
...
val pipeline = new Pipeline().setStages(Array(document_assembler, sentence_detector, tokenizer, word_embeddings, ner, ner_converter))
val data = Seq("Keratinocyte growth factor and acidic fibroblast growth factor are mitogens for primary cultures of mammary epithelium.").toDF("text")
val result = pipeline.fit(data).transform(data)

import nlu
nlu.load("en.med_ner.chemprot.clinical").predict("""Keratinocyte growth factor and acidic fibroblast growth factor are mitogens for primary cultures of mammary epithelium.""")

Results

+----+---------------------------------+---------+-------+----------+
|    | chunk                           |   begin |   end | entity   |
+====+=================================+=========+=======+==========+
|  0 | Keratinocyte growth factor      |       0 |    25 | GENE-Y   |
+----+---------------------------------+---------+-------+----------+
|  1 | acidic fibroblast growth factor |      31 |    61 | GENE-Y   |
+----+---------------------------------+---------+-------+----------+

Model Information

Model Name: ner_chemprot_clinical
Type: ner
Compatibility: Healthcare NLP 2.6.0 +
Edition: Official
License: Licensed
Input Labels: [sentence, token, embeddings]
Output Labels: [ner]
Language: [en]
Case sensitive: false

Data Source

This model was trained on the ChemProt corpus using ‘embeddings_clinical’ embeddings. Make sure you use the same embeddings when running the model.

Benchmarking

|    | label         |     tp |    fp |   fn |     prec |      rec |       f1 |
|---:|:--------------|-------:|------:|-----:|---------:|---------:|---------:|
|  0 | B-GENE-Y      |   4650 |  1090 |  838 | 0.810105 | 0.847303 | 0.828286 |
|  1 | B-GENE-N      |   1732 |   981 | 1019 | 0.638408 | 0.629589 | 0.633968 |
|  2 | I-GENE-Y      |   1846 |   571 |  573 | 0.763757 | 0.763125 | 0.763441 |
|  3 | B-CHEMICAL    |   7512 |   804 | 1136 | 0.903319 | 0.86864  | 0.88564  |
|  4 | I-CHEMICAL    |   1059 |   169 |  253 | 0.862378 | 0.807165 | 0.833858 |
|  5 | I-GENE-N      |   1393 |   853 |  598 | 0.620214 | 0.699648 | 0.657541 |
|  6 | Macro-average | 18192  | 4468  | 4417 | 0.766363 | 0.769245 | 0.767801 |
|  7 | Micro-average | 18192  | 4468  | 4417 | 0.802824 | 0.804635 | 0.803729 |