Description
Pretrained named entity recognition deep learning model for clinical terms. The SparkNLP deep learning model (NerDL) is inspired by a former state of the art model for NER: Chiu & Nicols, Named Entity Recognition with Bidirectional LSTM-CNN.
Predicted Entities
PROBLEM
, TEST
, TREATMENT
.
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.
document_assembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = SentenceDetector()\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
word_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")\
.setInputCols(["sentence", "token"])\
.setOutputCol("embeddings")
clinical_ner = MedicalNerModel.pretrained("ner_clinical", "en", "clinical/models") \
.setInputCols(["sentence", "token", "embeddings"]) \
.setOutputCol("ner")
...
nlpPipeline = Pipeline(stages=[document_assembler, sentence_detector, tokenizer, word_embeddings, clinical_ner, ner_converter])
model = nlpPipeline.fit(spark.createDataFrame([[""]]).toDF("text"))
results = model.transform(spark.createDataFrame([["The patient is a 21-day-old Caucasian male here for 2 days of congestion - mom has been suctioning yellow discharge from the patient's nares, plus she has noticed some mild problems with his breathing while feeding (but negative for any perioral cyanosis or retractions). One day ago, mom also noticed a tactile temperature and gave the patient Tylenol. Baby also has had some decreased p.o. intake. His normal breast-feeding is down from 20 minutes q.2h. to 5 to 10 minutes secondary to his respiratory congestion. He sleeps well, but has been more tired and has been fussy over the past 2 days. The parents noticed no improvement with albuterol treatments given in the ER. His urine output has also decreased; normally he has 8 to 10 wet and 5 dirty diapers per 24 hours, now he has down to 4 wet diapers per 24 hours. Mom denies any diarrhea. His bowel movements are yellow colored and soft in nature."]], ["text"]))
val document_assembler = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val sentence_detector = new SentenceDetector()
.setInputCols("document")
.setOutputCol("sentence")
val tokenizer = new Tokenizer()
.setInputCols("sentence")
.setOutputCol("token")
val word_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")
.setInputCols(Array("sentence", "token"))
.setOutputCol("embeddings")
val ner = MedicalNerModel.pretrained("ner_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("The patient is a 21-day-old Caucasian male here for 2 days of congestion - mom has been suctioning yellow discharge from the patient's nares, plus she has noticed some mild problems with his breathing while feeding (but negative for any perioral cyanosis or retractions). One day ago, mom also noticed a tactile temperature and gave the patient Tylenol. Baby also has had some decreased p.o. intake. His normal breast-feeding is down from 20 minutes q.2h. to 5 to 10 minutes secondary to his respiratory congestion. He sleeps well, but has been more tired and has been fussy over the past 2 days. The parents noticed no improvement with albuterol treatments given in the ER. His urine output has also decreased; normally he has 8 to 10 wet and 5 dirty diapers per 24 hours, now he has down to 4 wet diapers per 24 hours. Mom denies any diarrhea. His bowel movements are yellow colored and soft in nature.").toDF("text")
val result = pipeline.fit(data).transform(data)
Results
The output is a dataframe with a sentence per row and a "ner"
column containing all of the entity labels in the sentence, entity character indices, and other metadata. To get only the tokens and entity labels, without the metadata, select "token.result"
and "ner.result"
from your output dataframe or add the "Finisher"
to the end of your pipeline.
+-------------------------------------+---------+
|chunk |ner |
+-------------------------------------+---------+
|congestion |PROBLEM |
|suctioning yellow discharge |PROBLEM |
|some mild problems with his breathing|PROBLEM |
|any perioral cyanosis |PROBLEM |
|retractions |PROBLEM |
|a tactile temperature |TEST |
|Tylenol |TREATMENT|
|his respiratory congestion |PROBLEM |
|more tired |PROBLEM |
|fussy |PROBLEM |
|albuterol treatments |TREATMENT|
|His urine output |TEST |
|dirty diapers |TREATMENT|
|diarrhea |PROBLEM |
|yellow colored |PROBLEM |
+-------------------------------------+---------+
Model Information
Model Name: | ner_clinical |
Type: | ner |
Compatibility: | Spark NLP 3.0.0+ |
Edition: | Official |
License: | Licensed |
Input Labels: | [sentence,token, embeddings] |
Output Labels: | [ner] |
Language: | [en] |
Case sensitive: | false |
Data Source
Trained with augmented version of i2b2 dataset with embeddings_clinical
.
https://portal.dbmi.hms.harvard.edu/projects/n2c2-nlp/
Benchmarking
label tp fp fn prec rec f1
I-TREATMENT 6492 873 1445 0.881466 0.817941 0.848517
I-PROBLEM 15645 1808 2031 0.896408 0.885098 0.890717
B-PROBLEM 11160 1048 1424 0.914155 0.88684 0.90029
I-TEST 6878 864 1132 0.888401 0.858677 0.873286
B-TEST 8140 932 1081 0.897266 0.882768 0.889958
B-TREATMENT 8163 945 1150 0.896245 0.876517 0.886271
Macro-average 56478 6470 8263 0.895657 0.867974 0.881598
Micro-average 56478 6470 8263 0.897217 0.872368 0.884618