Description
This SDOH NER model is designed to detect and label social determinants of health (SDOH) demographic entities within text data. Social determinants of health are crucial factors that influence individuals’ health outcomes, encompassing various social, economic, and environmental elements. The model has been trained using advanced machine-learning techniques on a diverse range of text sources. The model’s accuracy and precision have been carefully validated against expert-labeled data to ensure reliable and consistent results. Here are the labels of the SDOH NER model with their description:
Age: All mention of ages including “Newborn, Infant, Child, Teenager, Teenage, Adult, etc.”Family_Member: Nouns that refer to a family member. “mother, father, brother, sister, etc.”Gender: Gender-specific nouns and pronounsGeographic_Entity: Geographical location refers to a specific physical point on Earth.Language: A system of conventional spoken, manual (signed), or written symbols by means of which human beings express themselves. “English, Spanish-speaking, etc. “Race_Ethnicity: The race and ethnicity categories include racial, ethnic, and national origins.Spiritual_Beliefs: Spirituality is concerned with beliefs beyond self, usually related to the existence of a superior being. “spiritual beliefs, religious beliefs, etc.”
Predicted Entities
Age, Family_Member, Gender, Geographic_Entity, Language, Race_Ethnicity, Spiritual_Beliefs
Live Demo Open in Colab Copy S3 URI
How to use
from pyspark.sql.types import StringType
document_assembler = DocumentAssembler()\
.setInputCol("text")\
.setOutputCol("document")
sentence_detector = SentenceDetectorDLModel.pretrained("sentence_detector_dl", "en")\
.setInputCols(["document"])\
.setOutputCol("sentence")
tokenizer = Tokenizer()\
.setInputCols(["sentence"])\
.setOutputCol("token")
clinical_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")\
.setInputCols(["sentence", "token"])\
.setOutputCol("embeddings")
ner_model = MedicalNerModel.pretrained("ner_sdoh_demographics", "en", "clinical/models")\
.setInputCols(["sentence", "token", "embeddings"])\
.setOutputCol("ner")
ner_converter = NerConverterInternal()\
.setInputCols(["sentence", "token", "ner"])\
.setOutputCol("ner_chunk")
pipeline = Pipeline(stages=[
document_assembler,
sentence_detector,
tokenizer,
clinical_embeddings,
ner_model,
ner_converter
])
sample_texts = ["Maria, a 35-year-old Hispanic woman, visited her healthcare provider for a routine check-up. As a bilingual individual, she was able to communicate comfortably in both English and Spanish. Maria shared that she practices Catholicism and finds strength and comfort in her faith. She has religious beliefs.",
"Ali, a 20 years old African American man, lives in New York. He is Muslim. He lives with his parents.",
"SOCIAL HISTORY: He is a former tailor from Korea. He has a Buddhist faith. He is an older adult.",
"He lives alone in France, single and no children. He speaks French and Portuguese. He is a member of Protestant Church",
"Pt is at her 60s married, Caucasian, Orthodox woman. Pt speaks English reasonably well. She lives with her son and husband."]
data = spark.createDataFrame(sample_texts, StringType()).toDF("text")
result = pipeline.fit(data).transform(data)
val document_assembler = new DocumentAssembler()
.setInputCol("text")
.setOutputCol("document")
val sentence_detector = SentenceDetectorDLModel.pretrained("sentence_detector_dl", "en")
.setInputCols("document")
.setOutputCol("sentence")
val tokenizer = new Tokenizer()
.setInputCols("sentence")
.setOutputCol("token")
val clinical_embeddings = WordEmbeddingsModel.pretrained("embeddings_clinical", "en", "clinical/models")
.setInputCols(Array("sentence", "token"))
.setOutputCol("embeddings")
val ner_model = MedicalNerModel.pretrained("ner_sdoh_demographics", "en", "clinical/models")
.setInputCols(Array("sentence", "token", "embeddings"))
.setOutputCol("ner")
val ner_converter = new NerConverterInternal()
.setInputCols(Array("sentence", "token", "ner"))
.setOutputCol("ner_chunk")
val pipeline = new Pipeline().setStages(Array(
document_assembler,
sentence_detector,
tokenizer,
clinical_embeddings,
ner_model,
ner_converter
))
val data = Seq(Array("Maria, a 35-year-old Hispanic woman, visited her healthcare provider for a routine check-up. As a bilingual individual, she was able to communicate comfortably in both English and Spanish. Maria shared that she practices Catholicism and finds strength and comfort in her faith. She has religious beliefs.",
"Ali, a 20 years old African American man, lives in New York. He is Muslim. He lives with his parents.",
"SOCIAL HISTORY: He is a former tailor from Korea. He has a Buddhist faith. He is an older adult.",
"He lives alone in France, single and no children. He speaks French and Portuguese. He is a member of Protestant Church",
"Pt is at her 60s married, Caucasian, Orthodox woman. Pt speaks English reasonably well. She lives with her son and husband.")).toDS.toDF("text")
val result = pipeline.fit(data).transform(data)
Results
+-----------------+-----+---+-----------------+
|chunk |begin|end|ner_label |
+-----------------+-----+---+-----------------+
|35-year-old |9 |19 |Age |
|Hispanic |21 |28 |Race_Ethnicity |
|woman |30 |34 |Gender |
|her |45 |47 |Gender |
|bilingual |98 |106|Language |
|she |120 |122|Gender |
|English |168 |174|Language |
|Spanish |180 |186|Language |
|she |207 |209|Gender |
|Catholicism |221 |231|Spiritual_Beliefs|
|her |267 |269|Gender |
|faith |271 |275|Spiritual_Beliefs|
|She |278 |280|Gender |
|religious beliefs|286 |302|Spiritual_Beliefs|
|20 years old |7 |18 |Age |
|African American |20 |35 |Race_Ethnicity |
|man |37 |39 |Gender |
|New York |51 |58 |Geographic_Entity|
|He |61 |62 |Gender |
|Muslim |67 |72 |Spiritual_Beliefs|
|He |75 |76 |Gender |
|his |89 |91 |Gender |
|parents |93 |99 |Family_Member |
|He |16 |17 |Gender |
|Korea |43 |47 |Geographic_Entity|
|He |50 |51 |Gender |
|Buddhist faith |59 |72 |Spiritual_Beliefs|
|He |75 |76 |Gender |
|older adult |84 |94 |Age |
|He |0 |1 |Gender |
|France |18 |23 |Geographic_Entity|
|children |40 |47 |Family_Member |
|He |50 |51 |Gender |
|French |60 |65 |Language |
|Portuguese |71 |80 |Language |
|He |83 |84 |Gender |
|Protestant |101 |110|Spiritual_Beliefs|
|her |9 |11 |Gender |
|60s |13 |15 |Age |
|Caucasian |26 |34 |Race_Ethnicity |
|Orthodox |37 |44 |Spiritual_Beliefs|
|woman |46 |50 |Gender |
|English |63 |69 |Language |
|She |88 |90 |Gender |
|her |103 |105|Gender |
|son |107 |109|Family_Member |
|husband |115 |121|Family_Member |
+-----------------+-----+---+-----------------+
Model Information
| Model Name: | ner_sdoh_demographics |
| Compatibility: | Healthcare NLP 4.4.4+ |
| License: | Licensed |
| Edition: | Official |
| Input Labels: | [sentence, token, embeddings] |
| Output Labels: | [ner] |
| Language: | en |
| Size: | 3.0 MB |
| Dependencies: | embeddings_clinical |
References
Internal SDOH Project
Benchmarking
label precision recall f1-score support
Age 0.93 0.94 0.93 470
Family_Member 0.98 0.99 0.99 1958
Gender 0.99 0.98 0.99 5021
Geographic_Entity 0.92 0.93 0.92 104
Language 1.00 0.91 0.95 22
Race_Ethnicity 1.00 0.96 0.98 28
Spiritual_Beliefs 0.96 0.91 0.94 57
micro-avg 0.98 0.98 0.98 7660
macro-avg 0.97 0.95 0.96 7660
weighted-avg 0.98 0.98 0.98 7660