Description
This model extracts biological and genetics terms in cancer-related texts using pre-trained NER model. This model is trained with the BertForTokenClassification method from the transformers library and imported into Spark NLP.
Predicted Entities
Amino_acid, Anatomical_system, Cancer, Cell, Cellular_component, Developing_anatomical_Structure, Gene_or_gene_product, Immaterial_anatomical_entity, Multi-tissue_structure, Organ, Organism, Organism_subdivision, Simple_chemical, Tissue, Organism_substance, Pathological_formation
How to use
documentAssembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
tokenizer = Tokenizer()\
.setInputCols(["document"])\
.setOutputCol("token")
tokenClassifier = BertForTokenClassification.pretrained("bert_token_classifier_ner_bionlp", "en", "clinical/models")\
.setInputCols("token", "document")\
.setOutputCol("ner")\
.setCaseSensitive(True)
ner_converter = NerConverter()\
.setInputCols(["document","token","ner"])\
.setOutputCol("ner_chunk")
pipeline = Pipeline(stages=[documentAssembler, tokenizer, tokenClassifier, ner_converter])
p_model = pipeline.fit(spark.createDataFrame(pd.DataFrame({'text': ['']})))
test_sentence = """Both the erbA IRES and the erbA/myb virus constructs transformed erythroid cells after infection of bone marrow or blastoderm cultures. The erbA/myb IRES virus exhibited a 5-10-fold higher transformed colony forming efficiency than the erbA IRES virus in the blastoderm assay."""
result = p_model.transform(spark.createDataFrame(pd.DataFrame({'text': [test_sentence]})))
val documentAssembler = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val tokenizer = new Tokenizer()
.setInputCols("document")
.setOutputCol("token")
val tokenClassifier = BertForTokenClassification.pretrained("bert_token_classifier_ner_bionlp", "en", "clinical/models")
.setInputCols(Array("token", "document"))
.setOutputCol("ner")
.setCaseSensitive(True)
val ner_converter = new NerConverter()
.setInputCols(Array("document","token","ner"))
.setOutputCol("ner_chunk")
val pipeline = new Pipeline().setStages(Array(documentAssembler, tokenizer, tokenClassifier, ner_converter))
val data = Seq("""Both the erbA IRES and the erbA/myb virus constructs transformed erythroid cells after infection of bone marrow or blastoderm cultures. The erbA/myb IRES virus exhibited a 5-10-fold higher transformed colony forming efficiency than the erbA IRES virus in the blastoderm assay.""").toDS.toDF("text")
val result = pipeline.fit(data).transform(data)
import nlu
nlu.load("en.classify.token_bert.bionlp").predict("""Both the erbA IRES and the erbA/myb virus constructs transformed erythroid cells after infection of bone marrow or blastoderm cultures. The erbA/myb IRES virus exhibited a 5-10-fold higher transformed colony forming efficiency than the erbA IRES virus in the blastoderm assay.""")
Results
+-------------------+----------------------+
|chunk |ner_label |
+-------------------+----------------------+
|erbA IRES |Organism |
|erbA/myb virus |Organism |
|erythroid cells |Cell |
|bone marrow |Multi-tissue_structure|
|blastoderm cultures|Cell |
|erbA/myb IRES virus|Organism |
|erbA IRES virus |Organism |
|blastoderm |Cell |
+-------------------+----------------------+
Model Information
| Model Name: | bert_token_classifier_ner_bionlp |
| Compatibility: | Healthcare NLP 3.3.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence, token] |
| Output Labels: | [ner] |
| Language: | en |
| Case sensitive: | true |
| Max sentense length: | 256 |
Data Source
Trained on Cancer Genetics (CG) task of the BioNLP Shared Task 2013. https://aclanthology.org/W13-2008/
Benchmarking
label precision recall f1-score support
B-Cancer 0.88 0.82 0.85 924
B-Cell 0.84 0.86 0.85 1013
B-Cellular_component 0.87 0.84 0.86 180
B-Developing_anatomical_structure 0.65 0.65 0.65 17
B-Gene_or_gene_product 0.62 0.79 0.69 2520
B-Immaterial_anatomical_entity 0.68 0.74 0.71 31
B-Multi-tissue_structure 0.84 0.76 0.80 303
B-Organ 0.78 0.74 0.76 156
B-Organism 0.93 0.86 0.89 518
B-Organism_subdivision 0.74 0.51 0.61 39
B-Organism_substance 0.93 0.66 0.77 102
B-Pathological_formation 0.85 0.60 0.71 88
B-Simple_chemical 0.61 0.75 0.68 727
B-Tissue 0.74 0.83 0.78 184
I-Amino_acid 0.60 1.00 0.75 3
I-Cancer 0.91 0.69 0.78 604
I-Cell 0.98 0.74 0.84 1091
I-Cellular_component 0.88 0.62 0.73 69
I-Multi-tissue_structure 0.89 0.86 0.87 162
I-Organ 0.67 0.59 0.62 17
I-Organism 0.84 0.45 0.59 120
I-Organism_substance 0.80 0.50 0.62 24
I-Pathological_formation 0.81 0.56 0.67 39
I-Tissue 0.83 0.86 0.84 111
accuracy - - 0.64 12129
macro-avg 0.73 0.56 0.60 12129
weighted-avg 0.83 0.64 0.68 12129