Detect Problem Entities (PROBLEM)

Description

This pipeline can be used to extracts problem (diseases, disorders, injuries, symptoms, signs .etc) information in medical text.

Predicted Entities

PROBLEM

Copy S3 URI

How to use

from sparknlp.pretrained import PretrainedPipeline

ner_pipeline = PretrainedPipeline("ner_problem_benchmark_pipeline", "en", "clinical/models")

text = """HISTORY OF PRESENT ILLNESS :
Mr. He is a 77 year old male with squamous cell carcinoma of the lung .
Over the pat three to four weeks , he started having increased dyspnea and noted wheezing .
A bronchoscopy showed protrusion of the tumor into the right main stem bronchus with a positive needle biopsy , washings and brushings for squamous cell carcinoma .
A computerized tomography scan showed a large subcarinal mass .
PAST MEDICAL HISTORY :
His past medical history was significant for malignant bladder tumor in 1991 .
PHYSICAL EXAMINATION :
On physical examination , Mr. He had very marked inspiratory and expiratory stridor .
There were no nodes present .
The breath sounds were somewhat decreased through both lung fields .
His cardiac examination did not show any murmur , gallop , or cardiomegaly .
There was no hepatosplenomegaly , and no peripheral edema .
HOSPITAL COURSE :
A bronchoscopy with the intention of coring out tumor was carried out by Dr. Reg He , but all the tumor was extrinsic to the airway and he was unable to relieve the obstruction .
The tumor now involves the trachea as well as the right main bronchus .
His major complaint was of persistent severe coughing and secretions .
"""

result = ner_pipeline.fullAnnotate(text)
from sparknlp.pretrained import PretrainedPipeline

ner_pipeline = nlp.PretrainedPipeline("ner_problem_benchmark_pipeline", "en", "clinical/models")

text = """HISTORY OF PRESENT ILLNESS :
Mr. He is a 77 year old male with squamous cell carcinoma of the lung .
Over the pat three to four weeks , he started having increased dyspnea and noted wheezing .
A bronchoscopy showed protrusion of the tumor into the right main stem bronchus with a positive needle biopsy , washings and brushings for squamous cell carcinoma .
A computerized tomography scan showed a large subcarinal mass .
PAST MEDICAL HISTORY :
His past medical history was significant for malignant bladder tumor in 1991 .
PHYSICAL EXAMINATION :
On physical examination , Mr. He had very marked inspiratory and expiratory stridor .
There were no nodes present .
The breath sounds were somewhat decreased through both lung fields .
His cardiac examination did not show any murmur , gallop , or cardiomegaly .
There was no hepatosplenomegaly , and no peripheral edema .
HOSPITAL COURSE :
A bronchoscopy with the intention of coring out tumor was carried out by Dr. Reg He , but all the tumor was extrinsic to the airway and he was unable to relieve the obstruction .
The tumor now involves the trachea as well as the right main bronchus .
His major complaint was of persistent severe coughing and secretions .
"""

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

val ner_pipeline = PretrainedPipeline("ner_problem_benchmark_pipeline", "en", "clinical/models")

val text = """HISTORY OF PRESENT ILLNESS :
Mr. He is a 77 year old male with squamous cell carcinoma of the lung .
Over the pat three to four weeks , he started having increased dyspnea and noted wheezing .
A bronchoscopy showed protrusion of the tumor into the right main stem bronchus with a positive needle biopsy , washings and brushings for squamous cell carcinoma .
A computerized tomography scan showed a large subcarinal mass .
PAST MEDICAL HISTORY :
His past medical history was significant for malignant bladder tumor in 1991 .
PHYSICAL EXAMINATION :
On physical examination , Mr. He had very marked inspiratory and expiratory stridor .
There were no nodes present .
The breath sounds were somewhat decreased through both lung fields .
His cardiac examination did not show any murmur , gallop , or cardiomegaly .
There was no hepatosplenomegaly , and no peripheral edema .
HOSPITAL COURSE :
A bronchoscopy with the intention of coring out tumor was carried out by Dr. Reg He , but all the tumor was extrinsic to the airway and he was unable to relieve the obstruction .
The tumor now involves the trachea as well as the right main bronchus .
His major complaint was of persistent severe coughing and secretions .
"""

val result = ner_pipeline.fullAnnotate(text)

Results

|    | chunk                   |   begin |   end | ner_label   |
|---:|:------------------------|--------:|------:|:------------|
|  0 | squamous cell carcinoma |      63 |    85 | PROBLEM     |
|  1 | dyspnea                 |     164 |   170 | PROBLEM     |
|  2 | wheezing                |     182 |   189 | PROBLEM     |
|  3 | tumor                   |     233 |   237 | PROBLEM     |
|  4 | squamous cell carcinoma |     332 |   354 | PROBLEM     |
|  5 | mass                    |     415 |   418 | PROBLEM     |
|  6 | malignant bladder tumor |     490 |   512 | PROBLEM     |
|  7 | murmur                  |     773 |   778 | PROBLEM     |
|  8 | cardiomegaly            |     794 |   805 | PROBLEM     |
|  9 | hepatosplenomegaly      |     822 |   839 | PROBLEM     |
| 10 | peripheral edema        |     850 |   865 | PROBLEM     |
| 11 | tumor                   |     935 |   939 | PROBLEM     |
| 12 | tumor                   |     985 |   989 | PROBLEM     |
| 13 | obstruction             |    1052 |  1062 | PROBLEM     |
| 14 | tumor                   |    1070 |  1074 | PROBLEM     |
| 15 | coughing                |    1183 |  1190 | PROBLEM     |

Model Information

Model Name: ner_problem_benchmark_pipeline
Type: pipeline
Compatibility: Healthcare NLP 5.5.3+
License: Licensed
Edition: Official
Language: en
Size: 1.7 GB

Included Models

  • DocumentAssembler
  • SentenceDetector
  • TokenizerModel
  • WordEmbeddingsModel
  • TextMatcherInternalModel
  • TextMatcherInternalModel
  • MedicalNerModel
  • NerConverterInternalModel
  • MedicalNerModel
  • NerConverterInternalModel
  • ChunkMergeModel
  • ChunkMergeModel

Benchmarking

       label  precision    recall  f1-score   support
           O      0.989     0.996     0.993     76426
     PROBLEM      0.948     0.866     0.905      6145
    accuracy      -         -         0.986     82571
   macro-avg      0.969     0.931     0.949     82571
weighted-avg      0.986     0.986     0.986     82571