Description
This relation extraction model links Date and Relative_Date extractions to clinical entities such as Test or Cancer_Dx.
Predicted Entities
is_date_of, O
How to use
Each relevant relation pair in the pipeline should include one date entity (Date or Relative_Date) and a clinical entity (such as Pathology_Test, Cancer_Dx or Chemotherapy).
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 = NerConverterInternal() \
.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(["Cancer_Dx-Date", "Date-Cancer_Dx", "Relative_Date-Cancer_Dx", "Cancer_Dx-Relative_Date", "Cancer_Surgery-Date", "Date-Cancer_Surgery", "Cancer_Surgery-Relative_Date", "Relative_Date-Cancer_Surgery"])
re_model = RelationExtractionDLModel.pretrained("redl_oncology_temporal_biobert", "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([["Her breast cancer was diagnosed last year."]]).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("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_oncology_wip", "en", "clinical/models")
.setInputCols(Array("sentence", "token", "embeddings"))
.setOutputCol("ner")
val ner_converter = new NerConverterInternal()
.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(Array("ner_chunk", "dependencies"))
.setOutputCol("re_ner_chunk")
.setMaxSyntacticDistance(10)
.setRelationPairs(Array("Cancer_Dx-Date", "Date-Cancer_Dx", "Relative_Date-Cancer_Dx", "Cancer_Dx-Relative_Date", "Cancer_Surgery-Date", "Date-Cancer_Surgery", "Cancer_Surgery-Relative_Date", "Relative_Date-Cancer_Surgery"))
val re_model = RelationExtractionDLModel.pretrained("redl_oncology_temporal_biobert", "en", "clinical/models")
.setPredictionThreshold(0.5f)
.setInputCols(Array("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("Her breast cancer was diagnosed last year.").toDS.toDF("text")
val result = pipeline.fit(data).transform(data)
Results
+----------+---------+-------------+-----------+-------------+-------------+-------------+-----------+---------+----------+
| relation| entity1|entity1_begin|entity1_end| chunk1| entity2|entity2_begin|entity2_end| chunk2|confidence|
+----------+---------+-------------+-----------+-------------+-------------+-------------+-----------+---------+----------+
|is_date_of|Cancer_Dx| 4| 16|breast cancer|Relative_Date| 32| 40|last year| 0.9999256|
+----------+---------+-------------+-----------+-------------+-------------+-------------+-----------+---------+----------+
Model Information
| Model Name: | redl_oncology_temporal_biobert |
| Compatibility: | Healthcare NLP 5.4.0+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [re_ner_chunk, sentence] |
| Output Labels: | [relation_extraction] |
| Language: | en |
| Size: | 405.4 MB |
References
In-house annotated oncology case reports.
Benchmarking
label recall precision f1 support
O 0.77 0.81 0.79 302.0
is_date_of 0.82 0.78 0.80 298.0
macro-avg 0.79 0.79 0.79 -