Description
Identify if tests were conducted on a particular date or any diagnosis was made on a specific date by checking relations between clinical entities and dates. 1 : Shows date and the clinical entity are related, 0 : Shows date and the clinical entity are not related.
Predicted Entities
0, 1
Live Demo Open in Colab Copy S3 URI
How to use
...
documenter = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentencer = SentenceDetector()\
.setInputCols(["document"])\
.setOutputCol("sentences")
tokenizer = sparknlp.annotators.Tokenizer()\
.setInputCols(["sentences"])\
.setOutputCol("tokens")
pos_tagger = PerceptronModel()\
.pretrained("pos_clinical", "en", "clinical/models") \
.setInputCols(["sentences", "tokens"])\
.setOutputCol("pos_tags")
words_embedder = WordEmbeddingsModel() \
.pretrained("embeddings_clinical", "en", "clinical/models") \
.setInputCols(["sentences", "tokens"]) \
.setOutputCol("embeddings")
ner_tagger = MedicalNerModel.pretrained("jsl_ner_wip_greedy_clinical", "en", "clinical/models")\
.setInputCols("sentences", "tokens", "embeddings")\
.setOutputCol("ner_tags")
ner_converter = NerConverter() \
.setInputCols(["sentences", "tokens", "ner_tags"]) \
.setOutputCol("ner_chunks")
dependency_parser = DependencyParserModel() \
.pretrained("dependency_conllu", "en") \
.setInputCols(["sentences", "pos_tags", "tokens"]) \
.setOutputCol("dependencies")
# Set a filter on pairs of named entities which will be treated as relation candidates
re_ner_chunk_filter = RENerChunksFilter() \
.setInputCols(["ner_chunks", "dependencies"])\
.setMaxSyntacticDistance(10)\
.setOutputCol("re_ner_chunks")\
.setRelationPairs(['symptom-date', 'date-procedure', 'relativedate-test', 'test-date'])
# The dataset this model is trained to is sentence-wise.
# This model can also be trained on document-level relations - in which case, while predicting, use "document" instead of "sentence" as input.
re_model = RelationExtractionDLModel()\
.pretrained('redl_date_clinical_biobert', 'en', "clinical/models") \
.setPredictionThreshold(0.5)\
.setInputCols(["re_ner_chunks", "sentences"]) \
.setOutputCol("relations")
pipeline = Pipeline(stages=[documenter, sentencer, tokenizer, pos_tagger, words_embedder, ner_tagger, ner_converter, dependency_parser, re_ner_chunk_filter, re_model])
text ="This 73 y/o patient had CT on 1/12/95, with progressive memory and cognitive decline since 8/11/94."
data = spark.createDataFrame([[text]]).toDF("text")
p_model = pipeline.fit(data)
result = p_model.transform(data)
...
val documenter = DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val sentencer = SentenceDetector()
.setInputCols("document")
.setOutputCol("sentences")
val tokenizer = sparknlp.annotators.Tokenizer()
.setInputCols("sentences")
.setOutputCol("tokens")
val pos_tagger = PerceptronModel()
.pretrained("pos_clinical", "en", "clinical/models")
.setInputCols(Array("sentences", "tokens"))
.setOutputCol("pos_tags")
val words_embedder = WordEmbeddingsModel()
.pretrained("embeddings_clinical", "en", "clinical/models")
.setInputCols(Array("sentences", "tokens"))
.setOutputCol("embeddings")
val ner_tagger = MedicalNerModel.pretrained("jsl_ner_wip_greedy_clinical", "en", "clinical/models")
.setInputCols(Array("sentences", "tokens", "embeddings"))
.setOutputCol("ner_tags")
val ner_converter = NerConverter()
.setInputCols(Array("sentences", "tokens", "ner_tags"))
.setOutputCol("ner_chunks")
val dependency_parser = DependencyParserModel()
.pretrained("dependency_conllu", "en")
.setInputCols(Array("sentences", "pos_tags", "tokens"))
.setOutputCol("dependencies")
// Set a filter on pairs of named entities which will be treated as relation candidates
val re_ner_chunk_filter = RENerChunksFilter()
.setInputCols(Array("ner_chunks", "dependencies"))
.setMaxSyntacticDistance(10)
.setOutputCol("re_ner_chunks")
.setRelationPairs(Array('symptom-date', 'date-procedure', 'relativedate-test', 'test-date'))
// The dataset this model is trained to is sentence-wise.
// This model can also be trained on document-level relations - in which case, while predicting, use "document" instead of "sentence" as input.
val re_model = RelationExtractionDLModel()
.pretrained("redl_date_clinical_biobert", "en", "clinical/models")
.setPredictionThreshold(0.5)
.setInputCols(Array("re_ner_chunks", "sentences"))
.setOutputCol("relations")
val pipeline = new Pipeline().setStages(Array(documenter, sentencer, tokenizer, pos_tagger, words_embedder, ner_tagger, ner_converter, dependency_parser, re_ner_chunk_filter, re_model))
val data = Seq("This 73 y/o patient had CT on 1/12/95, with progressive memory and cognitive decline since 8/11/94.").toDF("text")
val result = pipeline.fit(data).transform(data)
import nlu
nlu.load("en.relation.date").predict("""This 73 y/o patient had CT on 1/12/95, with progressive memory and cognitive decline since 8/11/94.""")
Results
| | relations | entity1 | entity1_begin | entity1_end | chunk1 | entity2 | entity2_end | entity2_end | chunk2 | confidence |
|---|-----------|---------|---------------|-------------|------------------------------------------|---------|-------------|-------------|---------|------------|
| 0 | 1 | Test | 24 | 25 | CT | Date | 31 | 37 | 1/12/95 | 1.0 |
| 1 | 1 | Symptom | 45 | 84 | progressive memory and cognitive decline | Date | 92 | 98 | 8/11/94 | 1.0 |
Model Information
| Model Name: | redl_date_clinical_biobert |
| Compatibility: | Healthcare NLP 2.7.3+ |
| License: | Licensed |
| Edition: | Official |
| Language: | en |
Data Source
Trained on an internal dataset.
Benchmarking
Relation Recall Precision F1 Support
0 0.738 0.729 0.734 84
1 0.945 0.947 0.946 416
Avg. 0.841 0.838 0.840