Pipeline to Detect Medication Entities, Assign Assertion Status and Find Relations

Description

A pipeline for detecting posology entities with the ner_posology_large NER model, assigning their assertion status with assertion_jsl model, and extracting relations between posology-related terminology with posology_re relation extraction model.

Predicted Entities

DRUG, DOSAGE, DURATION, FORM, FREQUENCY, ROUTE, STRENGTH

Download Copy S3 URI

How to use

from sparknlp.pretrained import PretrainedPipeline

pipeline = PretrainedPipeline("explain_clinical_doc_medication", "en", "clinical/models")

text = '''The patient is a 30-year-old female with a long history of insulin dependent diabetes, type 2. She received a course of Bactrim for 14 days for UTI.  She was prescribed 5000 units of Fragmin  subcutaneously daily, and along with Lantus 40 units subcutaneously at bedtime.'''

result = pipeline.fullAnnotate(text)
import com.johnsnowlabs.nlp.pretrained.PretrainedPipeline

val pipeline = new PretrainedPipeline("explain_clinical_doc_medication", "en", "clinical/models")

val text = "The patient is a 30-year-old female with a long history of insulin dependent diabetes, type 2. She received a course of Bactrim for 14 days for UTI.  She was prescribed 5000 units of Fragmin  subcutaneously daily, and along with Lantus 40 units subcutaneously at bedtime."

val result = pipeline.fullAnnotate(text)
import nlu
nlu.load("en.explain_dco.clinical_medication.pipeline").predict("""The patient is a 30-year-old female with a long history of insulin dependent diabetes, type 2. She received a course of Bactrim for 14 days for UTI.  She was prescribed 5000 units of Fragmin  subcutaneously daily, and along with Lantus 40 units subcutaneously at bedtime.""")

Results


+----+----------------+------------+
|    | chunks         | entities   |
|---:|:---------------|:-----------|
|  0 | insulin        | DRUG       |
|  1 | Bactrim        | DRUG       |
|  2 | for 14 days    | DURATION   |
|  3 | 5000 units     | DOSAGE     |
|  4 | Fragmin        | DRUG       |
|  5 | subcutaneously | ROUTE      |
|  6 | daily          | FREQUENCY  |
|  7 | Lantus         | DRUG       |
|  8 | 40 units       | DOSAGE     |
|  9 | subcutaneously | ROUTE      |
| 10 | at bedtime     | FREQUENCY  |
+----+----------------+------------+

+----+----------+------------+-------------+
|    | chunks   | entities   | assertion   |
|---:|:---------|:-----------|:------------|
|  0 | insulin  | DRUG       | Present     |
|  1 | Bactrim  | DRUG       | Past        |
|  2 | Fragmin  | DRUG       | Planned     |
|  3 | Lantus   | DRUG       | Planned     |
+----+----------+------------+-------------+

+----------------+-----------+------------+-----------+----------------+
| relation       | entity1   | chunk1     | entity2   | chunk2         |
|:---------------|:----------|:-----------|:----------|:---------------|
| DRUG-DURATION  | DRUG      | Bactrim    | DURATION  | for 14 days    |
| DOSAGE-DRUG    | DOSAGE    | 5000 units | DRUG      | Fragmin        |
| DRUG-ROUTE     | DRUG      | Fragmin    | ROUTE     | subcutaneously |
| DRUG-FREQUENCY | DRUG      | Fragmin    | FREQUENCY | daily          |
| DRUG-DOSAGE    | DRUG      | Lantus     | DOSAGE    | 40 units       |
| DRUG-ROUTE     | DRUG      | Lantus     | ROUTE     | subcutaneously |
| DRUG-FREQUENCY | DRUG      | Lantus     | FREQUENCY | at bedtime     |
+----------------+-----------+------------+-----------+----------------+

Model Information

Model Name: explain_clinical_doc_medication
Type: pipeline
Compatibility: Healthcare NLP 4.3.0+
License: Licensed
Edition: Official
Language: en
Size: 1.7 GB

Included Models

  • DocumentAssembler
  • SentenceDetectorDLModel
  • TokenizerModel
  • WordEmbeddingsModel
  • MedicalNerModel
  • NerConverterInternalModel
  • NerConverterInternalModel
  • AssertionDLModel
  • PerceptronModel
  • DependencyParserModel
  • PosologyREModel