Description
This model classifies the gender of a patient in a clinical document using context.
This model is a BioBERT-based classifier.
Predicted Entities
Female
, Male
, Unknown
How to use
document_assembler = DocumentAssembler() \
.setInputCol("text") \
.setOutputCol("document")
tokenizer = Tokenizer() \
.setInputCols(["document"]) \
.setOutputCol("token")
sequenceClassifier = MedicalBertForSequenceClassification.pretrained("bert_sequence_classifier_gender_biobert", "en", "clinical/models")\
.setInputCols(["document","token"]) \
.setOutputCol("class") \
.setCaseSensitive(True) \
.setMaxSentenceLength(512)
pipeline = Pipeline(stages=[
document_assembler,
tokenizer,
sequenceClassifier
])
data = spark.createDataFrame([["The patient took Advil and he experienced an adverse reaction."]]).toDF("text")
result = pipeline.fit(data).transform(data)
val documenter = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val tokenizer = new Tokenizer()
.setInputCols("sentences")
.setOutputCol("token")
val sequenceClassifier = MedicalBertForSequenceClassification.pretrained("bert_sequence_classifier_gender_biobert", "en", "clinical/models")
.setInputCols(Array("document","token"))
.setOutputCol("class")
val pipeline = new Pipeline().setStages(Array(documenter, tokenizer, sequenceClassifier))
val data = Seq("The patient took Advil and he experienced an adverse reaction.").toDF("text")
val result = pipeline.fit(data).transform(data)
import nlu
nlu.load("en.classify.gender.seq_biobert").predict("""The patient took Advil and he experienced an adverse reaction.""")
Results
+---------------------------------------------------------------+------+
|text |result|
+---------------------------------------------------------------+------+
|The patient took Advil and he experienced an adverse reaction. |[Male]|
+---------------------------------------------------------------+------+
Model Information
Model Name: | bert_sequence_classifier_gender_biobert |
Compatibility: | Healthcare NLP 3.4.1+ |
License: | Licensed |
Edition: | Official |
Input Labels: | [document, token] |
Output Labels: | [class] |
Language: | en |
Size: | 406.0 MB |
Case sensitive: | true |
Max sentence length: | 128 |
References
This model is trained on more than four thousands clinical documents (radiology reports, pathology reports, clinical visits, etc) annotated internally.
Benchmarking
label precision recall f1-score support
Female 0.94 0.94 0.94 479
Male 0.88 0.86 0.87 245
Unknown 0.73 0.78 0.76 102
accuracy 0.89 0.89 0.89 826
macro-avg 0.85 0.86 0.85 826
weighted-avg 0.90 0.89 0.90 826