Description
Pretrained named entity recognition deep learning model for clinical terms. The SparkNLP deep learning model (MedicalNerModel) is inspired by a former state-of-the-art model for NER: Chiu & Nicols, Named Entity Recognition with Bidirectional LSTM-CNN. This model is the version of ner_clinical_large model augmented with langtest library.
| test_type | before fail_count | after fail_count | before pass_count | after pass_count | minimum pass_rate | before pass_rate | after pass_rate |
|---|---|---|---|---|---|---|---|
| add_ocr_typo | 575 | 198 | 538 | 915 | 80% | 48% | 82% |
| titlecase | 530 | 353 | 913 | 1090 | 70% | 63% | 76% |
| uppercase | 841 | 331 | 547 | 1057 | 70% | 39% | 76% |
| weighted average | 1946 | 882 | 1998 | 3062 | 73% | 50.66% | 77.64% |
Predicted Entities
PROBLEM, TEST, TREATMENT
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")
word_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")\
.setInputCols(["sentence", "token"])\
.setOutputCol("embeddings")
clinical_ner = MedicalNerModel.pretrained("ner_clinical_large_langtest", "en", "clinical/models") \
.setInputCols(["sentence", "token", "embeddings"]) \
.setOutputCol("ner")
ner_converter = NerConverter()\
.setInputCols(["sentence", "token", "ner"])\
.setOutputCol("ner_chunk")
nlpPipeline = Pipeline(stages=[document_assembler,
sentence_detector,
tokenizer,
word_embeddings,
clinical_ner,
ner_converter])
model = nlpPipeline.fit(spark.createDataFrame([[""]]).toDF("text"))
data = spark.createDataFrame([["""A 28-year-old female with a history of gestational diabetes mellitus diagnosed eight years prior to presentation and subsequent type two diabetes mellitus (T2DM), one prior episode of HTG-induced pancreatitis three years prior to presentation, and associated with an acute hepatitis, presented with a one-week history of polyuria, poor appetite, and vomiting. She was on metformin, glipizide, and dapagliflozin for T2DM and atorvastatin and gemfibrozil for HTG. She had been on dapagliflozin for six months at the time of presentation. Physical examination on presentation was significant for dry oral mucosa ; significantly , her abdominal examination was benign with no tenderness, guarding, or rigidity. Pertinent laboratory findings on admission were: serum glucose 111 mg/dl, creatinine 0.4 mg/dL, triglycerides 508 mg/dL, total cholesterol 122 mg/dL, and venous pH 7.27."""]]).toDF("text")
result = model.transform(data)
val document_assembler = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val sentence_detector = SentenceDetectorDLModel.pretrained("sentence_detector_dl_healthcare", "en", "clinical/models")
.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_large_langtest", "en", "clinical/models")
.setInputCols("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, word_embeddings, ner, ner_converter))
val data = Seq("""A 28-year-old female with a history of gestational diabetes mellitus diagnosed eight years prior to presentation and subsequent type two diabetes mellitus (T2DM), one prior episode of HTG-induced pancreatitis three years prior to presentation, and associated with an acute hepatitis, presented with a one-week history of polyuria, poor appetite, and vomiting. She was on metformin, glipizide, and dapagliflozin for T2DM and atorvastatin and gemfibrozil for HTG. She had been on dapagliflozin for six months at the time of presentation. Physical examination on presentation was significant for dry oral mucosa; significantly, her abdominal examination was benign with no tenderness, guarding, or rigidity. Pertinent laboratory findings on admission were: serum glucose 111 mg/dl, creatinine 0.4 mg/dL, triglycerides 508 mg/dL, total cholesterol 122 mg/dL, and venous pH 7.27.""").toDS().toDF("text")
val result = pipeline.fit(data).transform(data)
Results
+-------------------------------------+---------+
|chunk |ner_label|
+-------------------------------------+---------+
|gestational diabetes mellitus |PROBLEM |
|subsequent type two diabetes mellitus|PROBLEM |
|T2DM |PROBLEM |
|HTG-induced pancreatitis |PROBLEM |
|an acute hepatitis |PROBLEM |
|polyuria |PROBLEM |
|poor appetite |PROBLEM |
|vomiting |PROBLEM |
|metformin |TREATMENT|
|glipizide |TREATMENT|
|dapagliflozin |TREATMENT|
|T2DM |PROBLEM |
|atorvastatin |TREATMENT|
|gemfibrozil |TREATMENT|
|HTG |PROBLEM |
|dapagliflozin |TREATMENT|
|Physical examination |TEST |
|dry oral mucosa |PROBLEM |
|her abdominal examination |TEST |
|tenderness |PROBLEM |
|guarding |PROBLEM |
|rigidity |PROBLEM |
|serum glucose |TEST |
|creatinine |TEST |
|triglycerides |TEST |
|total cholesterol |TEST |
|venous pH |TEST |
+-------------------------------------+---------+
Model Information
| Model Name: | ner_clinical_large_langtest |
| Compatibility: | Healthcare NLP 5.1.1+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence, token, embeddings] |
| Output Labels: | [ner] |
| Language: | en |
| Size: | 14.5 MB |
References
Trained on augmented version of 2010 i2b2 challenge data with ‘embeddings_clinical’.
Benchmarking
label precision recall f1-score support
B-PROBLEM 0.87 0.86 0.87 1094
I-PROBLEM 0.85 0.86 0.85 1549
B-TEST 0.88 0.88 0.88 721
I-TEST 0.85 0.86 0.86 656
B-TREATMENT 0.86 0.86 0.86 701
I-TREATMENT 0.79 0.86 0.82 644
micro-avg 0.85 0.86 0.86 5365
macro-avg 0.85 0.86 0.86 5365
weighted-avg 0.85 0.86 0.86 5365