Description
This relation extraction model identifies relations between dates and other clinical entities, between tumor mentions and their size, between anatomical entities and other clinical entities, and between tests and their results. In contrast to re_oncology_granular, all these relation types are labeled as is_related_to. The different types of relations can be identified considering the pairs of entities that are linked.
Predicted Entities
is_related_to
Live Demo Open in Colab Copy S3 URI
How to use
Use relation pairs to include only the combinations of entities that are relevant in your case.
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")
ner = MedicalNerModel.pretrained("ner_oncology_wip", "en", "clinical/models") \
.setInputCols(["sentence", "token", "embeddings"]) \
.setOutputCol("ner")
ner_converter = NerConverter() \
.setInputCols(["sentence", "token", "ner"]) \
.setOutputCol("ner_chunk")
pos_tagger = PerceptronModel.pretrained("pos_clinical", "en", "clinical/models") \
.setInputCols(["sentence", "token"]) \
.setOutputCol("pos_tags")
dependency_parser = DependencyParserModel.pretrained("dependency_conllu", "en") \
.setInputCols(["sentence", "pos_tags", "token"]) \
.setOutputCol("dependencies")
re_ner_chunk_filter = RENerChunksFilter() .setInputCols(["ner_chunk", "dependencies"]) .setOutputCol("re_ner_chunk") .setMaxSyntacticDistance(10) .setRelationPairs(["Tumor_Finding-Tumor_Size", "Tumor_Size-Tumor_Finding", "Cancer_Surgery-Relative_Date", "Relative_Date-Cancer_Surgery"])
re_model = RelationExtractionDLModel.pretrained("redl_oncology_biobert_wip", "en", "clinical/models") .setInputCols(["re_ner_chunk", "sentence"]) .setOutputCol("relation_extraction")
pipeline = Pipeline(stages=[document_assembler,
sentence_detector,
tokenizer,
word_embeddings,
ner,
ner_converter,
pos_tagger,
dependency_parser,
re_ner_chunk_filter,
re_model])
data = spark.createDataFrame([["A mastectomy was performed two months ago, and a 3 cm mass was extracted."]]).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_healthcare","en","clinical/models")
.setInputCols(Array("document"))
.setOutputCol("sentence")
val tokenizer = new Tokenizer()
.setInputCols(Array("sentence"))
.setOutputCol("token")
val word_embeddings = WordEmbeddingsModel().pretrained("embeddings_clinical", "en", "clinical/models")
.setInputCols(Array("sentence", "token"))
.setOutputCol("embeddings")
val ner = MedicalNerModel.pretrained("ner_oncology_wip", "en", "clinical/models")
.setInputCols(Array("sentence", "token", "embeddings"))
.setOutputCol("ner")
val ner_converter = new NerConverter()
.setInputCols(Array("sentence", "token", "ner"))
.setOutputCol("ner_chunk")
val pos_tagger = PerceptronModel.pretrained("pos_clinical", "en", "clinical/models")
.setInputCols(Array("sentence", "token"))
.setOutputCol("pos_tags")
val dependency_parser = DependencyParserModel.pretrained("dependency_conllu", "en")
.setInputCols(Array("sentence", "pos_tags", "token"))
.setOutputCol("dependencies")
val re_ner_chunk_filter = new RENerChunksFilter()
.setInputCols("ner_chunk", "dependencies")
.setOutputCol("re_ner_chunk")
.setMaxSyntacticDistance(10)
.setRelationPairs(Array("Tumor_Finding-Tumor_Size", "Tumor_Size-Tumor_Finding", "Cancer_Surgery-Relative_Date", "Relative_Date-Cancer_Surgery"))
val re_model = RelationExtractionDLModel.pretrained("redl_oncology_biobert_wip", "en", "clinical/models")
.setPredictionThreshold(0.5f)
.setInputCols("re_ner_chunk", "sentence")
.setOutputCol("relation_extraction")
val pipeline = new Pipeline().setStages(Array(document_assembler,
sentence_detector,
tokenizer,
word_embeddings,
ner,
ner_converter,
pos_tagger,
dependency_parser,
re_ner_chunk_filter,
re_model))
val data = Seq("A mastectomy was performed two months ago, and a 3 cm mass was extracted.").toDS.toDF("text")
val result = pipeline.fit(data).transform(data)
import nlu
nlu.load("en.relation.oncology_biobert_wip").predict("""A mastectomy was performed two months ago, and a 3 cm mass was extracted.""")
Results
| chunk1 | entity1 | chunk2 | entity2 | relation | confidence |
|:-----------|:---------------|:---------------|:--------------|:--------------|-------------:|
| mastectomy | Cancer_Surgery | two months ago | Relative_Date | is_related_to | 0.914221 |
| 3 cm | Tumor_Size | mass | Tumor_Finding | is_related_to | 0.90399 |
Model Information
| Model Name: | redl_oncology_biobert_wip |
| Compatibility: | Healthcare NLP 4.1.0+ |
| License: | Licensed |
| Edition: | Official |
| Language: | en |
| Size: | 405.4 MB |
References
In-house annotated oncology case reports.
Benchmarking
label recall precision f1
O 0.82 0.89 0.86
is_related_to 0.90 0.84 0.87
macro-avg 0.86 0.87 0.86