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
Access_To_Care, Age, Alcohol, Childhood_Development, Diet, Disability, Eating_Disorder, Education, Employment, Environmental_Condition, Exercise, Family_Member, Financial_Status, Gender, Geographic_Entity, Healthcare_Institution, Housing, Hypertension, Income, Insurance_Status, Language, Legal_Issues, Marital_Status, Mental_Health, Obesity, Other_Disease, Other_SDoH_Keywords, Quality_Of_Life, Race_Ethnicity, Sexual_Activity, Sexual_Orientation, Smoking, Social_Exclusion, Social_Support, Spiritual_Beliefs, Transportation, Violence_Or_Abuse
How to use
document_assembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = SentenceDetector()\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
labels = [
'Access_To_Care', 'Age', 'Alcohol', 'Childhood_Development', 'Diet', 'Disability',
'Eating_Disorder', 'Education', 'Employment', 'Environmental_Condition', 'Exercise',
'Family_Member', 'Financial_Status', 'Gender', 'Geographic_Entity', 'Healthcare_Institution',
'Housing', 'Hypertension', 'Income', 'Insurance_Status', 'Language', 'Legal_Issues',
'Marital_Status', 'Mental_Health', 'Obesity', 'Other_Disease', 'Other_SDoH_Keywords',
'Quality_Of_Life', 'Race_Ethnicity', 'Sexual_Activity', 'Sexual_Orientation', 'Smoking',
'Social_Exclusion', 'Social_Support', 'Spiritual_Beliefs', 'Transportation', 'Violence_Or_Abuse'
]
pretrained_zero_shot_ner = PretrainedZeroShotNER().pretrained("zeroshot_ner_sdoh_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([["""Smith is 55 years old, living in New York, a divorced Mexcian American woman with financial problems. She speaks Spanish and Portuguese. She lives in an apartment. She has been struggling with diabetes for the past 10 years and has recently been experiencing frequent hospitalizations due to uncontrolled blood sugar levels. Smith works as a cleaning assistant and cannot access health insurance or paid sick leave. She has a son, a student at college. Pt with likely long-standing depression. She is aware she needs rehab. Pt reports having her catholic faith as a means of support as well. She has a long history of alcohol, beginning in her teens. She reports she has been a daily drinker for 30 years, most recently drinking beer daily. She smokes a pack of cigarettes a day."""]]).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 = [
'Access_To_Care', 'Age', 'Alcohol', 'Childhood_Development', 'Diet', 'Disability',
'Eating_Disorder', 'Education', 'Employment', 'Environmental_Condition', 'Exercise',
'Family_Member', 'Financial_Status', 'Gender', 'Geographic_Entity', 'Healthcare_Institution',
'Housing', 'Hypertension', 'Income', 'Insurance_Status', 'Language', 'Legal_Issues',
'Marital_Status', 'Mental_Health', 'Obesity', 'Other_Disease', 'Other_SDoH_Keywords',
'Quality_Of_Life', 'Race_Ethnicity', 'Sexual_Activity', 'Sexual_Orientation', 'Smoking',
'Social_Exclusion', 'Social_Support', 'Spiritual_Beliefs', 'Transportation', 'Violence_Or_Abuse'
]
pretrained_zero_shot_ner = medical.PretrainedZeroShotNER().pretrained("zeroshot_ner_sdoh_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([["""Smith is 55 years old, living in New York, a divorced Mexcian American woman with financial problems. She speaks Spanish and Portuguese. She lives in an apartment. She has been struggling with diabetes for the past 10 years and has recently been experiencing frequent hospitalizations due to uncontrolled blood sugar levels. Smith works as a cleaning assistant and cannot access health insurance or paid sick leave. She has a son, a student at college. Pt with likely long-standing depression. She is aware she needs rehab. Pt reports having her catholic faith as a means of support as well. She has a long history of alcohol, beginning in her teens. She reports she has been a daily drinker for 30 years, most recently drinking beer daily. She smokes a pack of cigarettes a day."""]]).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("Access_To_Care", "Age", "Alcohol", "Childhood_Development", "Diet", "Disability",
"Eating_Disorder", "Education", "Employment", "Environmental_Condition", "Exercise",
"Family_Member", "Financial_Status", "Gender", "Geographic_Entity", "Healthcare_Institution",
"Housing", "Hypertension", "Income", "Insurance_Status", "Language", "Legal_Issues",
"Marital_Status", "Mental_Health", "Obesity", "Other_Disease", "Other_SDoH_Keywords",
"Quality_Of_Life", "Race_Ethnicity", "Sexual_Activity", "Sexual_Orientation", "Smoking",
"Social_Exclusion", "Social_Support", "Spiritual_Beliefs", "Transportation", "Violence_Or_Abuse")
val pretrained_zero_shot_ner = PretrainedZeroShotNER().pretrained("zeroshot_ner_sdoh_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([["""Smith is 55 years old, living in New York, a divorced Mexcian American woman with financial problems. She speaks Spanish and Portuguese. She lives in an apartment. She has been struggling with diabetes for the past 10 years and has recently been experiencing frequent hospitalizations due to uncontrolled blood sugar levels. Smith works as a cleaning assistant and cannot access health insurance or paid sick leave. She has a son, a student at college. Pt with likely long-standing depression. She is aware she needs rehab. Pt reports having her catholic faith as a means of support as well. She has a long history of alcohol, beginning in her teens. She reports she has been a daily drinker for 30 years, most recently drinking beer daily. She smokes a pack of cigarettes a day."""]]).toDF("text")
val result = pipeline.fit(data).transform(data)
Results
+------------------+-----+---+-------------------+----------+
|chunk |begin|end|ner_label |confidence|
+------------------+-----+---+-------------------+----------+
|55 years old |9 |20 |Age |0.985943 |
|New York |33 |40 |Geographic_Entity |0.93693453|
|divorced |45 |52 |Marital_Status |0.9977914 |
|Mexcian American |54 |69 |Race_Ethnicity |0.7580686 |
|woman |71 |75 |Gender |0.99670666|
|financial problems|82 |99 |Financial_Status |0.98300755|
|She |102 |104|Gender |0.99510634|
|Spanish |113 |119|Language |0.98981714|
|Portuguese |125 |134|Language |0.9745797 |
|She |137 |139|Gender |0.99644274|
|apartment |153 |161|Housing |0.97819704|
|She |164 |166|Gender |0.997265 |
|diabetes |193 |200|Other_Disease |0.95706624|
|hospitalizations |268 |283|Other_SDoH_Keywords|0.69986516|
|cleaning assistant|342 |359|Employment |0.84834933|
|health insurance |379 |394|Insurance_Status |0.8337548 |
|She |416 |418|Gender |0.99722373|
|son |426 |428|Family_Member |0.99653184|
|student |433 |439|Education |0.53395987|
|college |444 |450|Education |0.6121527 |
|depression |482 |491|Mental_Health |0.9893406 |
|She |494 |496|Gender |0.99657154|
|she |507 |509|Gender |0.99869967|
|rehab |517 |521|Access_To_Care |0.9899094 |
|her |542 |544|Gender |0.9876253 |
|support |575 |581|Social_Support |0.97643244|
|She |593 |595|Gender |0.99524057|
|alcohol |619 |625|Alcohol |0.96861035|
|her |641 |643|Gender |0.9950836 |
|teens |645 |649|Age |0.92979825|
|She |652 |654|Gender |0.98894835|
|she |664 |666|Gender |0.9956169 |
|drinker |685 |691|Alcohol |0.8265905 |
|drinking |721 |728|Alcohol |0.7135196 |
|She |742 |744|Gender |0.99702305|
|smokes |746 |751|Smoking |0.7797317 |
|cigarettes |763 |772|Smoking |0.7294187 |
+------------------+-----+---+-------------------+----------+
Model Information
| Model Name: | zeroshot_ner_sdoh_large |
| Compatibility: | Healthcare NLP 5.5.1+ |
| License: | Licensed |
| Edition: | Official |
| Language: | en |
| Size: | 1.6 GB |
Benchmarking
label precision recall f1-score support
Access_To_Care 0.6424 0.8929 0.7472 1018
Age 0.8843 0.9547 0.9182 905
Alcohol 0.9254 0.9527 0.9389 677
Childhood_Development 0.4808 0.9615 0.6410 26
Diet 0.3397 0.7396 0.5656 96
Disability 0.8081 0.9412 0.8696 85
Eating_Disorder 0.8571 0.9600 0.9057 50
Education 0.6981 0.9098 0.7900 122
Employment 0.8755 0.9489 0.9107 4269
Environmental_Condition 0.3977 0.8293 0.5375 82
Exercise 0.5568 0.9810 0.7103 105
Family_Member 0.9758 0.9785 0.9771 4042
Financial_Status 0.5668 0.8131 0.6679 214
Gender 0.9931 0.9802 0.9866 10248
Geographic_Entity 0.7679 0.7544 0.7611 228
Healthcare_Institution 0.9062 0.3753 0.5308 1391
Housing 0.7224 0.8847 0.7953 850
Hypertension 0.4919 1.0000 0.6595 61
Income 0.6574 0.8256 0.7320 86
Insurance_Status 0.7326 0.7925 0.7613 159
Language 0.5286 0.9737 0.6852 38
Legal_Issues 0.3495 0.8279 0.4915 122
Marital_Status 0.9022 1.0000 0.9486 166
Mental_Health 0.6068 0.8893 0.7214 1003
O 0.9884 0.9739 0.9811 169565
Obesity 0.5641 0.7857 0.6567 28
Other_Disease 0.7644 0.9214 0.8356 1285
Other_SDoH_Keywords 0.6272 0.8550 0.7236 545
Quality_Of_Life 0.3376 0.5519 0.5189 241
Race_Ethnicity 0.6235 0.9464 0.7518 56
Sexual_Activity 0.5506 0.9245 0.6901 53
Sexual_Orientation 0.8636 0.9744 0.9157 39
Smoking 0.9295 0.9797 0.9539 148
Social_Exclusion 0.3409 0.9184 0.4972 49
Social_Support 0.8700 0.8961 0.8829 1367
Spiritual_Beliefs 0.6850 0.7982 0.7373 109
Transportation 0.4195 0.8462 0.5609 117
Violence_Or_Abuse 0.3631 0.7974 0.4990 153
accuracy - - 0.9653 199798
macro-avg 0.6735 0.8773 0.7436 199798
weighted-avg 0.9715 0.9653 0.9668 199798