Description
Zero-shot Named Entity Recognition (NER) enables the identification of entities in text with minimal effort. By leveraging pre-trained language models and contextual understanding, zero-shot NER extends entity recognition capabilities to new domains and languages. While the model card includes default labels as examples, it is important to highlight that users are not limited to these labels. The model is designed to support any set of entity labels, allowing users to adapt it to their specific use cases. For best results, it is recommended to use labels that are conceptually similar to the provided defaults.
Predicted Entities
Adenopathy, Age, Biomarker, Biomarker_Result, Body_Part, Cancer_Dx, Cancer_Surgery,
Cycle_Count, Cycle_Day, Date, Death_Entit, Directio, Dosage, Duration, Frequency,
Gender, Grade, Histological_Type, Imaging_Test, Invasion, Metastasis, Oncogene, Pathology_Test,
Race_Ethnicity, Radiation_Dose, Relative_Date, Response_To_Treatment, Route, Smoking_Status,
Staging, Therapy, Tumor_Finding, Tumor_Size
How to use
document_assembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = SentenceDetector()\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
labels = ["Adenopathy", "Age","Biomarker","Biomarker_Result","Body_Part","Cancer_Dx","Cancer_Surgery",
"Cycle_Count","Cycle_Day","Date","Death_Entit","Directio","Dosage","Duration","Frequency",
"Gender","Grade","Histological_Type","Imaging_Test","Invasion","Metastasis","Oncogene","Pathology_Test",
"Race_Ethnicity","Radiation_Dose","Relative_Date","Response_To_Treatment","Route","Smoking_Status",
"Staging","Therapy","Tumor_Finding","Tumor_Size"]
pretrained_zero_shot_ner = PretrainedZeroShotNER().pretrained("zeroshot_ner_oncology_large", "en", "clinical/models")\
.setInputCols("sentence", "token")\
.setOutputCol("ner")\
.setPredictionThreshold(0.5)\
.setLabels(labels)
ner_converter = NerConverterInternal()\
.setInputCols("sentence", "token", "ner")\
.setOutputCol("ner_chunk")
pipeline = Pipeline().setStages([
document_assembler,
sentence_detector,
tokenizer,
pretrained_zero_shot_ner,
ner_converter
])
data = spark.createDataFrame([["""Two years ago, the patient presented with a tumor in her left breast and adenopathies. She was diagnosed with invasive ductal carcinoma. Last week she was also found to have a lung metastasis."""]]).toDF("text")
result = pipeline.fit(data).transform(data)
document_assembler = nlp.DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = nlp.SentenceDetector()\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = nlp.Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
labels = ["Adenopathy", "Age","Biomarker","Biomarker_Result","Body_Part","Cancer_Dx","Cancer_Surgery",
"Cycle_Count","Cycle_Day","Date","Death_Entit","Directio","Dosage","Duration","Frequency",
"Gender","Grade","Histological_Type","Imaging_Test","Invasion","Metastasis","Oncogene","Pathology_Test",
"Race_Ethnicity","Radiation_Dose","Relative_Date","Response_To_Treatment","Route","Smoking_Status",
"Staging","Therapy","Tumor_Finding","Tumor_Size"]
pretrained_zero_shot_ner = medical.PretrainedZeroShotNER().pretrained("zeroshot_ner_oncology_large", "en", "clinical/models")\
.setInputCols("sentence", "token")\
.setOutputCol("ner")\
.setPredictionThreshold(0.5)\
.setLabels(labels)
ner_converter = medical.NerConverterInternal()\
.setInputCols("sentence", "token", "ner")\
.setOutputCol("ner_chunk")
pipeline = nlp.Pipeline().setStages([
document_assembler,
sentence_detector,
tokenizer,
pretrained_zero_shot_ner,
ner_converter
])
data = spark.createDataFrame([["""Two years ago, the patient presented with a tumor in her left breast and adenopathies. She was diagnosed with invasive ductal carcinoma. Last week she was also found to have a lung metastasis."""]]).toDF("text")
result = pipeline.fit(data).transform(data)
val document_assembler = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val sentence_detector = new SentenceDetector()
.setInputCols("document")
.setOutputCol("sentence")
val tokenizer = new Tokenizer()
.setInputCols("sentence")
.setOutputCol("token")
labels = Array("Adenopathy", "Age","Biomarker","Biomarker_Result","Body_Part","Cancer_Dx","Cancer_Surgery",
"Cycle_Count","Cycle_Day","Date","Death_Entit","Directio","Dosage","Duration","Frequency",
"Gender","Grade","Histological_Type","Imaging_Test","Invasion","Metastasis","Oncogene","Pathology_Test",
"Race_Ethnicity","Radiation_Dose","Relative_Date","Response_To_Treatment","Route","Smoking_Status",
"Staging","Therapy","Tumor_Finding","Tumor_Size")
val pretrained_zero_shot_ner = PretrainedZeroShotNER().pretrained("zeroshot_ner_oncology_large", "en", "clinical/models")
.setInputCols(Array("sentence", "token"))
.setOutputCol("ner")
.setPredictionThreshold(0.5)
.setLabels(labels)
val ner_converter = new NerConverterInternal()
.setInputCols(Array("sentence", "token", "ner"))
.setOutputCol("ner_chunk")
val pipeline = new Pipeline().setStages(Array(
document_assembler,
sentence_detector,
tokenizer,
pretrained_zero_shot_ner,
ner_converter
))
val data = Seq([["""Two years ago, the patient presented with a tumor in her left breast and adenopathies. She was diagnosed with invasive ductal carcinoma. Last week she was also found to have a lung metastasis."""]]).toDF("text")
val result = pipeline.fit(data).transform(data)
Results
+-------------+-----+---+-----------------+----------+
|chunk |begin|end|ner_label |confidence|
+-------------+-----+---+-----------------+----------+
|Two years ago|1 |13 |Relative_Date |0.9848315 |
|tumor |45 |49 |Tumor_Finding |0.99484324|
|her |54 |56 |Gender |0.99652606|
|left |58 |61 |Direction |0.99791497|
|breast |63 |68 |Body_Part |0.9953009 |
|adenopathies |74 |85 |Adenopathy |0.9020297 |
|She |88 |90 |Gender |0.99673635|
|invasive |111 |118|Histological_Type|0.90829986|
|ductal |120 |125|Histological_Type|0.8250756 |
|carcinoma |127 |135|Cancer_Dx |0.97008 |
|Last week |138 |146|Date |0.9312243 |
|she |148 |150|Gender |0.9977864 |
|lung |177 |180|Body_Part |0.9972607 |
|metastasis |182 |191|Metastasis |0.9868492 |
+-------------+-----+---+-----------------+----------+
Model Information
| Model Name: | zeroshot_ner_oncology_large |
| Compatibility: | Healthcare NLP 5.5.1+ |
| License: | Licensed |
| Edition: | Official |
| Language: | en |
| Size: | 1.6 GB |
label precision recall f1-score support
Adenopathy 0.6077 0.8587 0.7117 92
Age 0.9561 0.9944 0.9749 1074
Biomarker 0.7034 0.8744 0.7796 1752
Biomarker_Result 0.5618 0.9026 0.6925 1632
Body_Part 0.7044 0.9203 0.7980 3540
Cancer_Dx 0.8526 0.8125 0.8321 1360
Cancer_Surgery 0.7790 0.7437 0.7609 749
Cycle_Count 0.7750 0.8857 0.8267 350
Cycle_Day 0.6928 0.8346 0.7571 254
Date 0.9469 0.9881 0.9671 921
Death_Entity 0.8293 0.9444 0.8831 36
Direction 0.7025 0.9279 0.7996 957
Dosage 0.8776 0.8389 0.8578 1111
Duration 0.6605 0.8855 0.7566 760
Frequency 0.7386 0.8604 0.7948 394
Gender 0.9802 0.9984 0.9892 1286
Grade 0.6317 0.8178 0.7128 258
Histological_Type 0.5459 0.7878 0.6449 476
Imaging_Test 0.8467 0.8653 0.8559 2145
Invasion 0.5461 0.8177 0.6549 181
Metastasis 0.9530 0.8756 0.9127 394
O 0.9709 0.8980 0.9330 66650
Oncogene 0.6892 0.6996 0.6944 466
Pathology_Test 0.6896 0.8382 0.7567 1100
Race_Ethnicity 0.8657 0.9831 0.9206 59
Radiation_Dose 0.8077 0.8936 0.8485 141
Relative_Date 0.7965 0.7710 0.7835 1284
Response_To_Treatment 0.5100 0.7988 0.6225 641
Route 0.6359 0.8493 0.7273 146
Smoking_Status 0.8983 0.9298 0.9138 57
Staging 0.5738 0.7848 0.6629 223
Therapy 0.7295 0.9021 0.8067 2012
Tumor_Finding 0.8879 0.8474 0.8672 1252
Tumor_Size 0.8069 0.9803 0.8852 1066
accuracy - - 0.8928 94819
macro avg 0.7575 0.8709 0.8054 94819
weighted avg 0.9096 0.8928 0.8975 94819