Few-Shot Assertion Model ( i2b2 )

Description

Assign assertion status to clinical entities extracted by NER based on their context in the text.

Predicted Entities

absent, associated_with_someone_else, conditional, hypothetical, possible, present

Copy S3 URI

How to use

document_assembler = DocumentAssembler()\
    .setInputCol("text")\
    .setOutputCol("document")

sentence_detector = SentenceDetector()\
    .setInputCols("document")\
    .setOutputCol("sentence")

tokenizer = Tokenizer()\
    .setInputCols(["sentence"])\
    .setOutputCol("token")

embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")\
    .setInputCols(["sentence", "token"])\
    .setOutputCol("embeddings")\
    .setCaseSensitive(False)

ner = MedicalNerModel.pretrained("ner_clinical", "en", "clinical/models")\
    .setInputCols(["sentence", "token", "embeddings"])\
    .setOutputCol("ner")

ner_converter = NerConverterInternal()\
    .setInputCols(["sentence", "token", "ner"])\
    .setOutputCol("ner_chunk")
    #.setWhiteList(["PROBLEM"])

few_shot_assertion_converter = FewShotAssertionSentenceConverter()\
    .setInputCols(["sentence", "token", "ner_chunk"])\
    .setOutputCol("assertion_sentence")

e5_embeddings = E5Embeddings\
    .pretrained("e5_base_v2_embeddings_medical_assertion_i2b2", "en", "clinical/models")\
    .setInputCols(["assertion_sentence"])\
    .setOutputCol("assertion_embedding")

few_shot_assertion_classifier = FewShotAssertionClassifierModel()\
    .pretrained("fewhot_assertion_i2b2_e5_base_v2_i2b2", "en", "clinical/models")\
    .setInputCols(["assertion_embedding"])\
    .setOutputCol("assertion")


pipeline = Pipeline(
    stages = [
        document_assembler,
        sentence_detector,
        tokenizer,
        embeddings,
        ner,
        ner_converter,
        few_shot_assertion_converter,
        e5_embeddings,
        few_shot_assertion_classifier
])

data = spark.createDataFrame([["""The patient is a 21-day-old Caucasian male here for 2 days of congestion - mom has been suctioning yellow discharge from the patient's nares, plus she has noticed some mild problems with his breathing while feeding (but negative for any perioral cyanosis or retractions). One day ago, mom also noticed a tactile temperature and gave the patient Tylenol. Baby also has had some decreased p.o. intake. His normal breast-feeding is down from 20 minutes q.2h. to 5 to 10 minutes secondary to his respiratory congestion. He sleeps well, but has been more tired and has been fussy over the past 2 days. The parents noticed no improvement with albuterol treatments given in the ER. His urine output has also decreased; normally he has 8 to 10 wet and 5 dirty diapers per 24 hours, now he has down to 4 wet diapers per 24 hours. Mom denies any diarrhea. His bowel movements are yellow colored and soft in nature."""]]).toDF("text")

result = resolver_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")

val embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")
    .setInputCols(Array("sentence", "token"))
    .setOutputCol("embeddings")
    .setCaseSensitive(False)

val ner = MedicalNerModel.pretrained("ner_clinical", "en", "clinical/models")
    .setInputCols(Array("sentence", "token", "embeddings"))
    .setOutputCol("ner")

val ner_converter = new NerConverterInternal()
    .setInputCols(Array("sentence", "token", "ner"))
    .setOutputCol("ner_chunk")
    //.setWhiteList("PROBLEM")

val few_shot_assertion_converter = new FewShotAssertionSentenceConverter()
    .setInputCols(Array("sentence","token","ner_chunk"))
    .setOutputCol("assertion_sentence")

val e5_embeddings = E5Embeddings
    .pretrained("e5_base_v2_embeddings_medical_assertion_i2b2", "en", "clinical/models")
    .setInputCols("assertion_sentence")
    .setOutputCol("assertion_embedding")

val few_shot_assertion_classifier = FewShotAssertionClassifierModel()
    .pretrained("fewhot_assertion_i2b2_e5_base_v2_i2b2", "en", "clinical/models")
    .setInputCols("assertion_embedding")
    .setOutputCol("assertion")


val pipeline = new Pipeline().setStages(Array(
        document_assembler,
        sentence_detector,
        tokenizer,
        embeddings,
        ner,
        ner_converter,
        few_shot_assertion_converter,
        e5_embeddings,
        few_shot_assertion_classifier
))

val data = Seq(Array("""The patient is a 21-day-old Caucasian male here for 2 days of congestion - mom has been suctioning yellow discharge from the patient's nares, plus she has noticed some mild problems with his breathing while feeding (but negative for any perioral cyanosis or retractions). One day ago, mom also noticed a tactile temperature and gave the patient Tylenol. Baby also has had some decreased p.o. intake. His normal breast-feeding is down from 20 minutes q.2h. to 5 to 10 minutes secondary to his respiratory congestion. He sleeps well, but has been more tired and has been fussy over the past 2 days. The parents noticed no improvement with albuterol treatments given in the ER. His urine output has also decreased; normally he has 8 to 10 wet and 5 dirty diapers per 24 hours, now he has down to 4 wet diapers per 24 hours. Mom denies any diarrhea. His bowel movements are yellow colored and soft in nature.""")).toDF("text")

val result = resolver_pipeline.fit(data).transform(data)

Results

| chunks                                |   begin |   end | entities   | assertion                    |   confidence |
|:--------------------------------------|--------:|------:|:-----------|:-----------------------------|-------------:|
| congestion                            |      63 |    72 | PROBLEM    | present                      |     0.952448 |
| suctioning yellow discharge           |      89 |   115 | TREATMENT  | associated_with_someone_else |     0.379236 |
| some mild problems with his breathing |     164 |   200 | PROBLEM    | present                      |     0.950967 |
| any perioral cyanosis                 |     234 |   254 | PROBLEM    | absent                       |     0.955528 |
| retractions                           |     259 |   269 | PROBLEM    | absent                       |     0.955448 |
| a tactile temperature                 |     303 |   323 | PROBLEM    | present                      |     0.954495 |
| Tylenol                               |     346 |   352 | TREATMENT  | present                      |     0.953851 |
| his respiratory congestion            |     489 |   514 | PROBLEM    | present                      |     0.952367 |
| more tired                            |     546 |   555 | PROBLEM    | present                      |     0.95374  |
| albuterol treatments                  |     638 |   657 | TREATMENT  | present                      |     0.954199 |
| His urine output                      |     676 |   691 | TEST       | present                      |     0.952049 |
| 4 wet diapers                         |     794 |   806 | TREATMENT  | present                      |     0.953411 |
| any diarrhea                          |     833 |   844 | PROBLEM    | absent                       |     0.954902 |
| yellow colored                        |     871 |   884 | PROBLEM    | present                      |     0.9547   |

Model Information

Model Name: fewhot_assertion_i2b2_e5_base_v2_i2b2
Compatibility: Healthcare NLP 5.3.3+
License: Licensed
Edition: Official
Input Labels: [assertion_embedding]
Output Labels: [assertion]
Language: en
Size: 25.4 KB

Benchmarking

                       label  precision    recall  f1-score   support
                      absent       0.94      0.97      0.95       303
associated_with_someone_else       0.94      0.88      0.91        17
                 conditional       0.60      0.20      0.30        15
                hypothetical       0.86      0.91      0.89        70
                    possible       0.65      0.85      0.74        60
                     present       0.96      0.94      0.95       880
                    accuracy          -         -      0.93      1345
                   macro-avg       0.83      0.79      0.79      1345
                weighted-avg       0.93      0.93      0.93      1345