Description
Using this relation extraction model, four relation types can be identified: is_date_of (between date entities and other clinical entities), is_size_of (between Tumor_Finding and Tumor_Size), is_location_of (between anatomical entities and other entities) and is_finding_of (between test entities and their results).
Predicted Entities
is_date_of, is_finding_of, is_location_of, is_size_of, O
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(['Date-Cancer_Dx',
'Tumor_Finding-Site_Breast',
'Tumor_Finding-Site_Bone',
'Tumor_Finding-Site_Liver',
'Tumor_Finding-Site_Lung',
'Tumor_Finding-Site_Lymph_Node',
'Tumor_Finding-Site_Other_Body_Part',
'Tumor_Fiding-Relative_Date',
'Tumor_Finding-Tumor_Size',
'Biomarker-Biomarker_Result',
'Pathology_Test-Cancer_Dx',
'Biomarker_Result-Biomarker',
'Imaging_Test-Tumor_Finding',
'Pathology_Test-Relative_Date',
'Pathology_Test-Pathology_Result',
'Relative_Date-Metastasis',
'Site-Lung-Metastasis',
'Tumor_Finding-Tumor_Size'
])
re_model = RelationExtractionDLModel.pretrained("redl_oncology_granular_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([["The Patient underwent a computed tomography scan, which showed a complex ovarian mass, 2 cm insize . A Pap smear performed one month later was positive for atypical glandular cells suspicious for adenocarcinoma. The pathologic specimen showed extension of the tumor throughout the fallopian tubes, appendix, omentum, and 5 out of 5 enlarged lymph nodes."]]).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(Array("ner_chunk","dependencies"))
.setOutputCol("re_ner_chunk")
.setMaxSyntacticDistance(10)
.setRelationPairs(Array(
"Date-Cancer_Dx",
"Tumor_Finding-Site_Breast",
"Tumor_Finding-Site_Bone",
"Tumor_Finding-Site_Liver",
"Tumor_Finding-Site_Lung",
"Tumor_Finding-Site_Lymph_Node",
"Tumor_Finding-Site_Other_Body_Part",
"Tumor_Fiding-Relative_Date",
"Tumor_Finding-Tumor_Size",
"Biomarker-Biomarker_Result",
"Pathology_Test-Cancer_Dx",
"Biomarker_Result-Biomarker",
"Imaging_Test-Tumor_Finding",
"Pathology_Test-Relative_Date",
"Pathology_Test-Pathology_Result",
"Relative_Date-Metastasis",
"Site-Lung-Metastasis",
"Tumor_Finding-Tumor_Size" ))
val re_model = RelationExtractionDLModel.pretrained("redl_oncology_granular_biobert_wip","en","clinical/models")
.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("The Patient underwent a computed tomography scan, which showed a complex ovarian mass, 2 cm insize . A Pap smear performed one month later was positive for atypical glandular cells suspicious for adenocarcinoma. The pathologic specimen showed extension of the tumor throughout the fallopian tubes, appendix, omentum, and 5 out of 5 enlarged lymph nodes.").toDF("text")
val result = pipeline.fit(data).transform(data)
import nlu
nlu.load("en.relation.oncology_granular_biobert_wip").predict("""The Patient underwent a computed tomography scan, which showed a complex ovarian mass, 2 cm insize . A Pap smear performed one month later was positive for atypical glandular cells suspicious for adenocarcinoma. The pathologic specimen showed extension of the tumor throughout the fallopian tubes, appendix, omentum, and 5 out of 5 enlarged lymph nodes.""")
Results
| | relation | entity1 | entity1_begin | entity1_end | chunk1 | entity2 | entity2_begin | entity2_end | chunk2 | confidence |
|--:|---------------:|---------------------:|--------------:|------------:|-------------------------:|---------------------:|--------------:|------------:|-------------------------:|-----------:|
| 0 | is_finding_of | Imaging_Test | 24 | 47 | computed tomography scan | Tumor_Finding | 81 | 84 | mass | 0.672964 |
| 1 | is_location_of | Site_Other_Body_Part | 73 | 79 | ovarian | Tumor_Finding | 81 | 84 | mass | 0.976508 |
| 2 | is_size_of | Tumor_Finding | 81 | 84 | mass | Tumor_Size | 87 | 90 | 2 cm | 0.952546 |
| 3 | is_date_of | Pathology_Test | 103 | 111 | Pap smear | Relative_Date | 123 | 137 | one month later | 0.927102 |
| 4 | is_finding_of | Pathology_Test | 103 | 111 | Pap smear | Pathology_Result | 156 | 179 | atypical glandular cells | 0.860861 |
| 5 | is_finding_of | Pathology_Test | 103 | 111 | Pap smear | Cancer_Dx | 196 | 209 | adenocarcinoma | 0.545740 |
| 6 | is_location_of | Tumor_Finding | 260 | 264 | tumor | Site_Other_Body_Part | 281 | 295 | fallopian tubes | 0.875905 |
| 7 | is_location_of | Tumor_Finding | 260 | 264 | tumor | Site_Other_Body_Part | 298 | 305 | appendix | 0.774170 |
| 8 | is_location_of | Tumor_Finding | 260 | 264 | tumor | Site_Other_Body_Part | 308 | 314 | omentum | 0.906041 |
Model Information
| Model Name: | redl_oncology_granular_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 0.83 0.91 0.87
is_date_of 0.82 0.80 0.81
is_finding_of 0.92 0.85 0.88
is_location_of 0.95 0.85 0.90
is_size_of 0.91 0.80 0.85
macro-avg 0.89 0.84 0.86