Edit on GitHub
  • John Snow Labs
  • Jul 25, 2022

Detect Anatomical Structures in Medical Text

  • en
  • ner
  • clinical
  • licensed
  • bertfortokenclassification

Description

Anatomical entities ranging from subcellular structures to organ systems are central to biomedical science, and mentions of these entities are essential to understanding the scientific literature.

This model is trained with the BertForTokenClassification method from the transformers library and imported into Spark NLP. The model detects anatomical structures from a medical text.

Predicted Entities

Anatomy

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")

ner_model = MedicalBertForTokenClassifier.pretrained("bert_token_classifier_ner_anatem", "en", "clinical/models")\
    .setInputCols(["sentence", "token"])\
    .setOutputCol("ner")\
    .setCaseSensitive(True)\
    .setMaxSentenceLength(512)

ner_converter = NerConverter()\
    .setInputCols(["sentence", "token", "ner"])\
    .setOutputCol("ner_chunk")

pipeline = Pipeline(stages=[
    document_assembler, 
    sentence_detector,
    tokenizer,
    ner_model,
    ner_converter   
    ])

data = spark.createDataFrame([["""Malignant cells often display defects in autophagy, an evolutionarily conserved pathway for degrading long-lived proteins and cytoplasmic organelles. However, as yet, there is no genetic evidence for a role of autophagy genes in tumor suppression. The beclin 1 autophagy gene is monoallelically deleted in 40 - 75 % of cases of human sporadic breast, ovarian, and prostate cancer."""]]).toDF("text")

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

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

val sentence_detector = SentenceDetectorDLModel.pretrained("sentence_detector_dl_healthcare", "en", "clinical/models")
    .setInputCols(Array("document"))
    .setOutputCol("sentence")

val tokenizer = new Tokenizer()
    .setInputCols(Array("sentence"))
    .setOutputCol("token")

val ner_model = MedicalBertForTokenClassifier.pretrained("bert_token_classifier_ner_anatem", "en", "clinical/models")
    .setInputCols(Array("sentence", "token"))
    .setOutputCol("ner")
    .setCaseSensitive(True)
    .setMaxSentenceLength(512)

val ner_converter = new NerConverter()
    .setInputCols(Array("sentence", "token", "ner"))
    .setOutputCol("ner_chunk")

val pipeline = new Pipeline().setStages(Array(document_assembler, 
                                                   sentence_detector,
                                                   tokenizer,
                                                   ner_model,
                                                   ner_converter))

val data = Seq("""Malignant cells often display defects in autophagy, an evolutionarily conserved pathway for degrading long-lived proteins and cytoplasmic organelles. However, as yet, there is no genetic evidence for a role of autophagy genes in tumor suppression. The beclin 1 autophagy gene is monoallelically deleted in 40 - 75 % of cases of human sporadic breast, ovarian, and prostate cancer.""").toDS.toDF("text")

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

import nlu
nlu.load("en.classify.token_bert.anatem").predict("""Malignant cells often display defects in autophagy, an evolutionarily conserved pathway for degrading long-lived proteins and cytoplasmic organelles. However, as yet, there is no genetic evidence for a role of autophagy genes in tumor suppression. The beclin 1 autophagy gene is monoallelically deleted in 40 - 75 % of cases of human sporadic breast, ovarian, and prostate cancer.""")

Results

+----------------------+-------+
|ner_chunk             |label  |
+----------------------+-------+
|Malignant cells       |Anatomy|
|cytoplasmic organelles|Anatomy|
|tumor                 |Anatomy|
|breast                |Anatomy|
|ovarian               |Anatomy|
|prostate cancer       |Anatomy|
+----------------------+-------+

Model Information

Model Name: bert_token_classifier_ner_anatem
Compatibility: Healthcare NLP 4.0.0+
License: Licensed
Edition: Official
Input Labels: [sentence, token]
Output Labels: [ner]
Language: en
Size: 404.2 MB
Case sensitive: true
Max sentence length: 512

References

https://github.com/cambridgeltl/MTL-Bioinformatics-2016

Benchmarking

 label         precision  recall  f1-score  support 
 B-Anatomy     0.8489     0.9380  0.8912    4616    
 I-Anatomy     0.9190     0.8839  0.9011    3247    
 micro-avg     0.8755     0.9157  0.8951    7863    
 macro-avg     0.8839     0.9110  0.8962    7863    
 weighted-avg  0.8778     0.9157  0.8953    7863