Detect Procedure Entities (PROCEDURE)

Description

This pipeline can be used to extract procedure mentions in medical text.

Predicted Entities

PROCEDURE

Copy S3 URI

How to use

from sparknlp.pretrained import PretrainedPipeline

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

text = """PRINCIPAL PROCEDURE :
4-2-93 , right and left heart catheterization ( transeptal ) with coronary graft and left ventriculogram .
4-8-93 , open heart aortic valve replacement and bypass of right coronary artery .

Mr. No was on the pump for 2 hours and 25 minutes , with an aortic crossclamp time of 1 hour and 48 minutes .
Postoperatively , the patient was extubated on the first postoperative day .
He had a good deal of pulmonary congestion .
He seemed to be doing well until the morning of 4-13-93 when he suddenly became pulseless .
There was no evidence of ventricular fibrillation .
Cardiopulmonary resuscitation was immediately undertaken , but was not successful .
The patient had his chest opened for any evidence of tamponade and there was no evidence of bleeding .
The heart appeared to be flaccid .
We really have no good explanation of what this was all about .
He was being treated for a possible pneumonia .
He did have a good deal of pulmonary congestion , but this occurred suddenly and unexpectedly .
Cardiopulmonary resuscitation efforts were carried out for virtually one hour , but were unsuccessful .
The patient was pronounced dead on the morning of 4-13-93 .
A post mortem examination will be performed .
"""

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

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

text = """PRINCIPAL PROCEDURE :
4-2-93 , right and left heart catheterization ( transeptal ) with coronary graft and left ventriculogram .
4-8-93 , open heart aortic valve replacement and bypass of right coronary artery .

Mr. No was on the pump for 2 hours and 25 minutes , with an aortic crossclamp time of 1 hour and 48 minutes .
Postoperatively , the patient was extubated on the first postoperative day .
He had a good deal of pulmonary congestion .
He seemed to be doing well until the morning of 4-13-93 when he suddenly became pulseless .
There was no evidence of ventricular fibrillation .
Cardiopulmonary resuscitation was immediately undertaken , but was not successful .
The patient had his chest opened for any evidence of tamponade and there was no evidence of bleeding .
The heart appeared to be flaccid .
We really have no good explanation of what this was all about .
He was being treated for a possible pneumonia .
He did have a good deal of pulmonary congestion , but this occurred suddenly and unexpectedly .
Cardiopulmonary resuscitation efforts were carried out for virtually one hour , but were unsuccessful .
The patient was pronounced dead on the morning of 4-13-93 .
A post mortem examination will be performed .
"""

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

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

val text = """PRINCIPAL PROCEDURE :
4-2-93 , right and left heart catheterization ( transeptal ) with coronary graft and left ventriculogram .
4-8-93 , open heart aortic valve replacement and bypass of right coronary artery .

Mr. No was on the pump for 2 hours and 25 minutes , with an aortic crossclamp time of 1 hour and 48 minutes .
Postoperatively , the patient was extubated on the first postoperative day .
He had a good deal of pulmonary congestion .
He seemed to be doing well until the morning of 4-13-93 when he suddenly became pulseless .
There was no evidence of ventricular fibrillation .
Cardiopulmonary resuscitation was immediately undertaken , but was not successful .
The patient had his chest opened for any evidence of tamponade and there was no evidence of bleeding .
The heart appeared to be flaccid .
We really have no good explanation of what this was all about .
He was being treated for a possible pneumonia .
He did have a good deal of pulmonary congestion , but this occurred suddenly and unexpectedly .
Cardiopulmonary resuscitation efforts were carried out for virtually one hour , but were unsuccessful .
The patient was pronounced dead on the morning of 4-13-93 .
A post mortem examination will be performed .
"""

val result = ner_pipeline.fullAnnotate(text)

Results

|    | chunk                               |   begin |   end | ner_label   |
|---:|:------------------------------------|--------:|------:|:------------|
|  0 | heart catheterization               |      46 |    66 | PROCEDURE   |
|  1 | transeptal                          |      70 |    79 | PROCEDURE   |
|  2 | coronary graft                      |      88 |   101 | PROCEDURE   |
|  3 | left ventriculogram                 |     107 |   125 | PROCEDURE   |
|  4 | open heart aortic valve replacement |     138 |   172 | PROCEDURE   |
|  5 | bypass                              |     178 |   183 | PROCEDURE   |
|  6 | aortic crossclamp                   |     273 |   289 | PROCEDURE   |
|  7 | extubated                           |     357 |   365 | PROCEDURE   |
|  8 | Cardiopulmonary resuscitation       |     589 |   617 | PROCEDURE   |
|  9 | Cardiopulmonary resuscitation       |    1019 |  1047 | PROCEDURE   |

Model Information

Model Name: ner_procedure_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
  • MedicalNerModel
  • NerConverterInternalModel
  • MedicalNerModel
  • NerConverterInternalModel
  • ChunkMergeModel
  • ChunkMergeModel

Benchmarking

       label  precision    recall  f1-score   support
           O      0.999     0.997     0.998     81024
   PROCEDURE      0.869     0.936     0.901      1547
    accuracy      -         -         0.996     82571
   macro-avg      0.934     0.967     0.950     82571
weighted-avg      0.996     0.996     0.996     82571