Edit on GitHub
  • John Snow Labs
  • Jun 22, 2022

Detect Living Species(roberta_embeddings_BR_BERTo)

  • pt
  • ner
  • clinical
  • licensed
  • roberta

Description

Extract living species from clinical texts in Portuguese which is critical to scientific disciplines like medicine, biology, ecology/biodiversity, nutrition and agriculture. This model is trained using roberta_embeddings_BR_BERTo embeddings.

It is trained on the LivingNER corpus that is composed of clinical case reports extracted from miscellaneous medical specialties including COVID, oncology, infectious diseases, tropical medicine, urology, pediatrics, and others.

NOTE :

  1. The text files were translated from Spanish with a neural machine translation system.
  2. The annotations were translated with the same neural machine translation system.
  3. The translated annotations were transferred to the translated text files using an annotation transfer technology.

Predicted Entities

HUMAN, SPECIES

Open in Colab Copy S3 URI

How to use

document_assembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")

sentence_detector = SentenceDetectorDLModel.pretrained("sentence_detector_dl", "xx")\
.setInputCols(["document"])\
.setOutputCol("sentence")

tokenizer = Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")

embeddings = RoBertaEmbeddings.pretrained("roberta_embeddings_BR_BERTo","pt")\
.setInputCols(["sentence", "token"])\
.setOutputCol("embeddings")

ner_model = MedicalNerModel.pretrained("ner_living_species_roberta", "pt", "clinical/models")\
.setInputCols(["sentence", "token", "embeddings"])\
.setOutputCol("ner")

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

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

data = spark.createDataFrame([["""Mulher de 23 anos, de Capinota, Cochabamba, Bolívia. Ela está no nosso país há quatro anos. Frequentou o departamento de emergência obstétrica onde foi encontrada grávida de 37 semanas, com um colo dilatado de 5 cm e membranas rompidas. O obstetra de emergência realizou um teste de estreptococos negativo e solicitou um hemograma, glucose, bioquímica básica, HBV, HCV e serologia da sífilis."""]]).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", "xx")
.setInputCols(Array("document"))
.setOutputCol("sentence")

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

val embeddings = RoBertaEmbeddings.pretrained("roberta_embeddings_BR_BERTo","pt")
.setInputCols(Array("sentence", "token"))
.setOutputCol("embeddings")

val ner_model = MedicalNerModel.pretrained("ner_living_species_roberta", "pt", "clinical/models")
.setInputCols(Array("sentence", "token", "embeddings"))
.setOutputCol("ner")

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

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

val data = Seq("""Mulher de 23 anos, de Capinota, Cochabamba, Bolívia. Ela está no nosso país há quatro anos. Frequentou o departamento de emergência obstétrica onde foi encontrada grávida de 37 semanas, com um colo dilatado de 5 cm e membranas rompidas. O obstetra de emergência realizou um teste de estreptococos negativo e solicitou um hemograma, glucose, bioquímica básica, HBV, HCV e serologia da sífilis.""").toDS.toDF("text")

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

import nlu
nlu.load("pt.med_ner.living_species.roberta").predict("""Mulher de 23 anos, de Capinota, Cochabamba, Bolívia. Ela está no nosso país há quatro anos. Frequentou o departamento de emergência obstétrica onde foi encontrada grávida de 37 semanas, com um colo dilatado de 5 cm e membranas rompidas. O obstetra de emergência realizou um teste de estreptococos negativo e solicitou um hemograma, glucose, bioquímica básica, HBV, HCV e serologia da sífilis.""")

Results

+-------------+-------+
|ner_chunk    |label  |
+-------------+-------+
|Mulher       |HUMAN  |
|grávida      |HUMAN  |
|estreptococos|SPECIES|
|HBV          |SPECIES|
|HCV          |SPECIES|
|sífilis      |SPECIES|
+-------------+-------+

Model Information

Model Name: ner_living_species_roberta
Compatibility: Healthcare NLP 3.5.3+
License: Licensed
Edition: Official
Input Labels: [sentence, token, embeddings]
Output Labels: [ner]
Language: pt
Size: 16.4 MB

References

https://temu.bsc.es/livingner/2022/05/03/multilingual-corpus/

Benchmarking

label         precision  recall  f1-score  support 
B-HUMAN       0.86       0.91    0.88      2827    
B-SPECIES     0.52       0.86    0.65      2796    
I-HUMAN       0.79       0.43    0.55      180     
I-SPECIES     0.62       0.81    0.70      1099    
micro-avg     0.65       0.86    0.74      6902    
macro-avg     0.69       0.75    0.70      6902    
weighted-avg  0.68       0.86    0.75      6902