Description
This model classifies related to the loss of social support such as a family member or friend in the clinical documents. A discharge summary was classified True for Community-Absent if the discharge summary had passages related to the loss of social support and False if such passages were not found in the discharge summary.
Predicted Entities
True, False
Live Demo Open in Colab Copy S3 URI
How to use
document_assembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
tokenizer = Tokenizer()\
.setInputCols("document")\
.setOutputCol("token")
sequenceClassifier = MedicalBertForSequenceClassification.pretrained("bert_sequence_classifier_sdoh_community_absent_status", "en", "clinical/models")\
.setInputCols(["document","token"])\
.setOutputCol("class")
pipeline = Pipeline(stages=[
document_assembler,
tokenizer,
sequenceClassifier
])
sample_texts =["She has two adult sons. She is a widow. She was employed with housework. She quit smoking 20 to 30 years ago, but smoked two packs per day for 20 to 30 years. She drinks one glass of wine occasionally. She avoids salt in her diet. ",
"65 year old male presented with several days of vice like chest pain. He states that he felt like his chest was being crushed from back to the front. Lives with spouse and two sons moved to US 1 month ago."]
data = spark.createDataFrame(sample_texts, StringType()).toDF("text")
result = pipeline.fit(data).transform(data)
val document_assembler = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val tokenizer = new Tokenizer()
.setInputCols("document")
.setOutputCol("token")
val sequenceClassifier = MedicalBertForSequenceClassification.pretrained("bert_sequence_classifier_sdoh_community_absent_status", "en", "clinical/models")
.setInputCols(Array("document","token"))
.setOutputCol("class")
val pipeline = new Pipeline().setStages(Array(document_assembler,
tokenizer,
sequenceClassifier))
val data = Seq("She has two adult sons. She is a widow. She was employed with housework. She quit smoking 20 to 30 years ago, but smoked two packs per day for 20 to 30 years. She drinks one glass of wine occasionally. She avoids salt in her diet.")
val result = pipeline.fit(data).transform(data)
import nlu
nlu.load("en.classify.bert_sequence.sdoh_community_absent_status").predict("""She has two adult sons. She is a widow. She was employed with housework. She quit smoking 20 to 30 years ago, but smoked two packs per day for 20 to 30 years. She drinks one glass of wine occasionally. She avoids salt in her diet. """)
Results
+----------------------------------------------------------------------------------------------------+-------+
| text| result|
+----------------------------------------------------------------------------------------------------+-------+
|She has two adult sons. She is a widow. She was employed with housework. She quit smoking 20 to 3...| [True]|
|65 year old male presented with several days of vice like chest pain. He states that he felt like...|[False]|
+----------------------------------------------------------------------------------------------------+-------+
Model Information
| Model Name: | bert_sequence_classifier_sdoh_community_absent_status |
| Compatibility: | Healthcare NLP 4.2.2+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [document, token] |
| Output Labels: | [class] |
| Language: | en |
| Size: | 410.9 MB |
| Case sensitive: | true |
| Max sentence length: | 512 |
Benchmarking
label precision recall f1-score support
False 0.89 0.77 0.83 155
True 0.63 0.80 0.70 74
accuracy - - 0.78 229
macro-avg 0.76 0.79 0.76 229
weighted-avg 0.80 0.78 0.79 229