Spark NLP for Healthcare Release Notes 5.3.1

5.3.1

Highlights

We are delighted to announce remarkable enhancements and updates in our latest release of Spark NLP for Healthcare. This release comes with MedDRA resolver, 2x faster optimized Deidentification pipelines, Response to Treatment classifier for oncology, and 41 new clinical pretrained models and pipelines. It’s as big as a major release!

• Welcoming MedDRA into the library. Releasing 8 new Entity Resolver and Mapper models to associate clinical entities with MedDRA LLT and PT codes.
• Enhancing assertion annotation workflow with AssertionMerger annotator to allow using multiple assertion models within the same pipeline.
• Adding new clinical deidentification pipelines that are 2x faster. Now we have 15 pretrained deidentification pipelines of various sizes and capabilities.
• Efficiency analysis and cost evaluation of deidentification pipelines on cloud platforms
• Updated Opioid NER model and Drug text matcher model
• New text classifier for Response to Treatment to detect response status/ outcome for the treatment applied for oncology patients.
• 2 new Entity Resolver models for associating SNOMED clinical entities.
• Clinical document analysis with one-liner pretrained pipelines for specific clinical tasks and concepts.
• A new augmented NER model for multilingual name extraction by leveraging the capabilities of the LangTest library to boost its robustness significantly.
• DatasetInfo parameter added to SentenceEntityResolver annotator to track the source datasets’ versions.
• Robust exception handling to allow skipping only the corrupted records processed via GenericClassifier, BertSentenceChunkEmbeddings, AssertionFilterer, ChunkFilterer, ContextualParser, ChunkMerge and Deidentification annotators.
• Changed the license of CPT and MedDRA models in the ModelHub, and attempting to use them in Healthcare NLP now throws an error.
• Various core improvements; bug fixes, enhanced overall robustness and reliability of Spark NLP for Healthcare
  • Fixed sentence positions in MedicalBertForSequenceClassification
  • Updated Deidentification Module according to the latest spark versions
  • Updated ALAB Module for assertion result according to tokenization flexibility
  • Deprecation of the setRel Method in ChunkMapper: Transitioning to the setRels parameter
  • Enhancements in SentenceEntityResolver: Bug Fix and Annotator Refactor
  • Added assertion_source, ner_chunk, and ner_label metadata fields to the AssertionDL and AssertionLogReg annotators
  • Implemented fixes and enhancements related to entity handling and resolution in Resolver and ChunkMapper, including incorporating an entity field in resolver metadata from embeddings, rectifying the entity field assignment in ChunkMapper, and resolving a bug with all_k_resolutions when using setMultivaluesRelations(True) in ChunkMapper
  • Updated init.py and the path of class for BertSentenceChunkEmbedding
• Updated notebooks and demonstrations for making Spark NLP for Healthcare easier to navigate and understand
  • New Task Based Clinical Pretrained Pipelines Notebook
  • Updated Clinical Assertion Model Notebook
  • Updated Snomed Entity Resolver Model Training Notebook
  • New Response to Treatment Classification Demo
  • Updated Opioid Demo
• The addition and update of numerous new clinical models and pipelines continue to reinforce our offering in the healthcare domain

These enhancements will elevate your experience with Spark NLP for Healthcare, enabling more efficient, accurate, and streamlined analysis of healthcare-related natural language data.

Welcoming MedDRA into the Library. Releasing 10 New Entity Resolver, Mapper Models and Pretrained Pipelines to Associate Clinical Entities With Meddra Llt and Pt Codes

Introducing 2 new Sentence Entity Resolver Models sbiobertresolve_meddra_lowest_level_term and sbiobertresolve_meddra_preferred_term help to map medical entities to MedDRA codes.

Example:

meddra_resolver = SentenceEntityResolverModel.load("sbiobertresolve_meddra_lowest_level_term") \
     .setInputCols(["sbert_embeddings"]) \
     .setOutputCol("meddra_llt_code")\
     .setDistanceFunction("EUCLIDEAN")

text= """This is an 82-year-old male with a history of prior tobacco use, benign hypertension, chronic renal insufficiency, chronic bronchitis, gastritis, and ischemic attack. He initially presented to Braintree with ST elevation and was transferred to St. Margaret’s Center. He underwent cardiac catheterization because of the left main coronary artery stenosis, which was complicated by hypotension and bradycardia. We describe the side effects of 5-FU in a colon cancer patient who suffered mucositis and dermatitis."""

Result:

• 6 ChunkMapper Models for Medical Code Mapping to Map Various Medical Terminologies Across Each Other

Introducing a suite of new ChunkMapper models designed to streamline medical code mapping tasks. These models include mappings between RxNorm, ICD-10, MedDRA-LLT, and MedDRA-PT codes, offering a comprehensive solution for interoperability within medical systems.

Example:

mapperModel = ChunkMapperModel.load('meddra_llt_pt_mapper')\
    .setInputCols(["ner_chunk"])\
    .setOutputCol("mappings")\
    .setRels(["icd10_code"])

text = ["10002442", "10000007", "10003696"]

Result:

• Introducing 2 New Pretrained Meddra Resolver Pipelines Designed For Effortless Integration With Just A Single Line Of Code

These pipelines are capable of extracting clinical entities and linking them to their respective MedDRA LLT and PT codes, while also facilitating mapping of these codes to LLT/PT or ICD-10 codes.

Example:

from sparknlp.pretrained import PretrainedPipeline

meddra_llt_pipeline = PretrainedPipeline.from_disk("meddra_llt_resolver_pipeline")

result = meddra_llt_pipeline.fullAnnotate('This is an 82-year-old male with a history of prior tobacco use, benign hypertension, chronic renal insufficiency, chronic bronchitis, gastritis, and ischemic attack. He initially presented to Braintree with ST elevation and was transferred to St. Margaret’s Center. He underwent cardiac catheterization because of the left main coronary artery stenosis, which was complicated by hypotension and bradycardia. We describe the side effects of 5-FU in a colon cancer patient who suffered mucositis and dermatitis.')

Result:

Important note: To utilize these MedDRA models/pipelines, possession of a valid MedDRA license is requisite. When you want to use these models and pipelines, you will receive a warning like below. If you possess a valid MedDRA license and wish to use this model, kindly contact us at support@johnsnowlabs.com.

IllegalArgumentException: 'meddra_llt_pt_mapper' model cannot be used as a pretrained model.
To load this model locally via .load(), possession of a valid MedDRA / CPT license is required.
If you possess one thru corresponding agencies and wish to use this model, contact us at support@johnsnowlabs.com.

Enhancing Assertion Annotation Workflow with AssertionMerger Annotator to Allow using Multiple Assertion Models within the Same Pipeline.

Introducing the latest addition to our annotation toolkit, the AssertionMerger Annotator, designed to streamline the merging process of assertion columns from various annotators like AssertionDL and AssertionLogReg. This powerful tool offers customizable parameters for filtering, prioritizing, and seamlessly combining assertion annotations. Learn how to leverage features like merging overlapping annotations, applying filters before or after merging, and prioritizing based on confidence levels and assertion sources. Optimize your annotation workflow with AssertionMerger Annotator, ensuring efficient and accurate consolidation of assertion data.

Parameters:

• mergeOverlapping: Whether to merge overlapping matched assertion annotations. Default: True
• applyFilterBeforeMerge: Whether to apply filtering before the merging process. If True, filtering will be applied before merging; if False, filtering will be applied after the merging process. Default: False.
• blackList: If defined, list of entities to ignore. The rest will be processed.
• whiteList: If defined, list of entities to process. The rest will be ignored. Do not include the IOB prefix on labels.
• caseSensitive: Determines whether the definitions of the white-listed and black-listed entities are case sensitive. Default: True.
• assertionsConfidence: Pairs (assertion, confidenceThreshold) to filter assertions that have confidence lower than the confidence threshold.
• orderingFeatures: Specifies the ordering features to use for overlapping entities. Possible values include: ‘begin’, ‘end’, ‘length’, ‘source’, and ‘confidence’. Default: ['begin', 'length', 'source']
• selectionStrategy: Determines the strategy for selecting annotations. Annotations can be selected either sequentially based on their order (Sequential) or using a more diverse strategy (DiverseLonger). Currently, only Sequential and DiverseLonger options are available. Default: Sequential.
• defaultConfidence: When the confidence value is included in the orderingFeatures and a given annotation does not have any confidence, this parameter determines the value to be used. The default value is 0.
• assertionSourcePrecedence: Specifies the assertion sources to use for prioritizing overlapping annotations when the ‘source’ ordering feature is utilized. This parameter contains a comma-separated list of assertion sources that drive the prioritization. Annotations will be prioritized based on the order of the given string.
• sortByBegin: Whether to sort the annotations by begin at the end of the merge and filter process. Default: False.

Example:

# Assertion model trained on i2b2 (sampled from MIMIC) dataset
assertion_jsl = AssertionDLModel.pretrained("assertion_jsl_augmented", "en", "clinical/models") \
    .setInputCols(["sentence", "ner_jsl_chunk", "embeddings"]) \
    .setOutputCol("assertion_jsl")\
    .setEntityAssertionCaseSensitive(False)

# Assertion model trained on radiology dataset
assertion_dl = AssertionDLModel.pretrained("assertion_dl", "en", "clinical/models") \
    .setInputCols(["sentence", "ner_clinical_chunk", "embeddings"]) \
    .setOutputCol("assertion_dl")

assertion_merger = AssertionMerger() \
    .setInputCols("assertion_jsl", "assertion_dl") \
    .setOutputCol("assertion_merger") \
    .setMergeOverlapping(True) \
    .setSelectionStrategy("sequential") \
    .setAssertionSourcePrecedence("assertion_dl, assertion_jsl") \
    .setCaseSensitive(False) \
    .setAssertionsConfidence({"past": 0.70}) \
    .setOrderingFeatures(["length", "source", "confidence"]) \
    .setDefaultConfidence(0.50)


text = [
    """Patient had a headache for the last 2 weeks, and appears anxious when she walks fast. No alopecia noted. She denies pain. Her father is paralyzed and it is a stressor for her. She got antidepressant. We prescribed sleeping pills for her current insomnia."""
]

Result:

Please check Clinical Assertion Model Notebook for more information

Adding New Clinical Deidentification Pipelines for Flexible Deployments

The Comprehensive Clinical Deidentification Pipeline offers a robust solution for anonymizing sensitive PHI (Protected Health Information) from medical texts. This versatile pipeline is equipped to mask and obfuscate a wide array of PHI entities including AAGE, CONTACT, DATE, ID, LOCATION, NAME, PROFESSION, CITY, COUNTRY, DOCTOR, HOSPITAL, IDNUM, MEDICALRECORD, ORGANIZATION, PATIENT, PHONE, PROFESSION, STREET, USERNAME, ZIP, ACCOUNT, LICENSE, VIN, SSN, DLN, PLATE, IPADDR, and more. With support for multiple languages including Arabic, German, French, English Spanish, Italian, Portuguese, and Romanian this pipeline ensures compliance with privacy regulations across diverse healthcare settings. Choose from various models optimized for different use cases, such as obfuscation levels and subentity recognition, to tailor the deidentification process according to specific requirements.

We will share a detailed table on our wiki page soon to explain the pros and cons of each model as well as tips and tricks to show how to use them effectively

Efficiency Analysis and Cost Evaluation of Deidentification Pipelines on Cloud Platforms

These results show speed benchmarks and cost evaluations for deidentification pipelines deployed across diverse cloud platforms, including AWS EMR and EC2. Additionally, forthcoming results from DataBricks promise to further enrich the analysis, offering deeper insights into de-identification pipeline performance.

Estimated Minimum Costs:

• EMR Base Pipeline: partition number: 256, 10K cost:$1.04, 1M cost:$104.41
• EMR Optimized Pipeline: partition number: 256, 10K cost:$0.54, 1M cost:$54.04
• EC2 Instance Base Pipeline: partition number: 512, 10K cost:$0.36, 1M cost:$35.70
• EC2 Instance Optimized Pipeline: partition number: 1024, 10K cost:$0.18, 1M cost:$17.85
• DataBricks results will be published soon.

Updated Opioid-related Named Entity Recognition and Drug-related Text Matcher Models

• ner_opioid: This Updated Opioid-related Named Entity Recognition model has been enhanced with new annotated text data. Opioids are a class of drugs that include the illegal drug heroin, synthetic opioids such as fentanyl, and pain relievers available legally by prescription. The model is designed to detect and label opioid-related entities within text data. It has been retrained using advanced deep learning techniques on an expanded and diversified range of text sources, including newly annotated text specifically focused on opioid-related content.

Example:

ner_model = MedicalNerModel.pretrained("ner_opioid", "en", "clinical/models")\
    .setInputCols(["sentence", "token", "embeddings"])\
    .setOutputCol("ner")

sample_texts = """History of Present Illness: A 20-year-old male was transferred from an outside hospital for evaluation for liver transplant following a Percocet overdose. On Sunday, March 27th, he experienced a stressful day and consumed approximately 20 Percocet (5/325) tablets throughout the day following a series of family arguments. He denies any intent to harm himself, although his parents confirm past suicidal attempts. On Monday, he felt he was experiencing a Percocet withdrawal "hangover" and took an additional 5 Percocet. He was admitted to the Surgical Intensive Care Unit (SICU) and received care from Liver, Transplant, Toxicology. Treatment included Naloxone every 4 hours, resulting in a gradual improvement in liver function tests (LFTs) and INR. During recovery, he developed hypertension and was initiated on clonidine."""

Result:

Benchmark:

                 label  precision    recall  f1-score   support
           alcohol_use       0.92      0.95      0.94       353
              antidote       1.00      0.99      0.99       141
  communicable_disease       0.76      0.88      0.82       224
         drug_duration       0.81      0.71      0.75       238
             drug_form       0.97      0.95      0.96       614
        drug_frequency       0.94      0.97      0.96      1527
         drug_quantity       0.96      0.94      0.95      2169
            drug_route       0.95      0.98      0.97       903
         drug_strength       0.84      0.95      0.89       388
            employment       0.79      0.63      0.70       306
      general_symptoms       0.90      0.84      0.87      4483
           legal_issue       0.73      0.52      0.61        84
        marital_status       0.95      0.95      0.95        57
           opioid_drug       0.98      0.96      0.97       725
         other_disease       0.91      0.90      0.90      4145
            other_drug       0.94      0.93      0.94      2617
     psychiatric_issue       0.88      0.85      0.86      1356
    sexual_orientation       1.00      0.78      0.88        23
substance_use_disorder       0.91      0.88      0.90       276
                  test       0.97      0.93      0.95       102
           test_result       1.00      0.93      0.97        30
              violence       0.81      0.71      0.76       542
             micro-avg       0.92      0.89      0.90     21303
             macro-avg       0.91      0.87      0.89     21303
          weighted-avg       0.91      0.89      0.90     21303

• drug_matcher: The latest iteration of the Drug-related Text Matcher Model has been enhanced significantly, boasting an expanded database with the inclusion of an additional 100 thousand drugs. Through meticulous curation, the model has undergone refinement by strategically eliminating words that may have previously led to false positives.

Example:

text_matcher = TextMatcherInternalModel.pretrained("drug_matcher","en","clinical/models") \
    .setInputCols(["sentence", "token"])\
    .setOutputCol("matched_text")\

sample_texts = """John's doctor prescribed aspirin for his heart condition, along with paracetamol for his fever and headache, amoxicillin for his tonsilitis and lansoprazole for his GORD on 2023-12-01."""

Result:

New Oncological Response to Treatment Classification Model

The Oncological Response to Treatment classifier was trained on a diverse dataset, this model provides accurate label assignments and confidence scores for its predictions. The primary goal of this model is to categorize text into two key labels: Yes and No.

Example:

sequenceClassifier = MedicalBertForSequenceClassification\
    .pretrained("bert_sequence_classifier_response_to_treatment", "en", "clinical/models")\
    .setInputCols(["document", "token"])\
    .setOutputCol("prediction")


sample_texts = [
    "The breast ultrasound after neoadjuvant chemotherapy displayed a decrease in the primary lesion size from 3 cm to 1 cm, suggesting a favorable response to treatment. The skin infection is also well controlled with a multi-antibiotic approach. ",
    "MRI of the pelvis indicated no further progression of endometriosis after laparoscopic excision and six months of hormonal suppression therapy.",
    "A repeat endoscopy revealed healing gastric ulcers with new signs of malignancy or H. pylori infection. Will discuss the PPI continuum.",
    "Dynamic contrast-enhanced MRI of the liver revealed no significant reduction in the size and number of hepatic metastases following six months of targeted therapy with sorafenib."
]

Result:

Benchmark:

       label  precision    recall  f1-score   support
          No     0.9927    0.9875    0.9901      3031
         Yes     0.8430    0.9027    0.8718       226
    accuracy          -         -    0.9816      3257
   macro-avg     0.9178    0.9451    0.9309      3257
weighted-avg     0.9823    0.9816    0.9819      3257

2 New Sentence Entity Resolver Models for Associating SNOMED Clinical Entities

Introducing 2 new Sentence Entity Resolver Models sbiobertresolve_snomed_no_class and sbiobertresolve_snomed_conditions help to map medical entities to SNOMED codes.

Example:

resolver = SentenceEntityResolverModel\
    .pretrained("sbiobertresolve_snomed_conditions", "en", "clinical/models")\
    .setInputCols(["sbert_embeddings"]) \
    .setOutputCol("resolution")\
    .setDistanceFunction("EUCLIDEAN")

text = """Medical professionals rushed in the bustling emergency room to attend to the patient with alarming symptoms.
The attending physician immediately noted signs of respiratory distress, including stridor, a high-pitched sound indicative of upper respiratory tract obstruction.
The patient, struggling to breathe, exhibited dyspnea. Concern heightened when they began experiencing syncope,
a sudden loss of consciousness likely stemming from inadequate oxygenation. Further examination revealed a respiratory tract hemorrhage."""

Result:

Clinical Document Analysis with One-Liner Pretrained Pipelines for Specific Clinical Tasks and Concepts

We introduce a suite of advanced, hybrid pretrained pipelines, specifically designed to streamline the process of analyzing clinical documents. These pipelines are built upon multiple state-of-the-art (SOTA) pretrained models, delivering a comprehensive solution for extracting vital information with unprecedented ease.

What sets this release apart is the elimination of complexities typically involved in building and chaining models. Users no longer need to navigate the intricacies of constructing intricate pipelines from scratch or the uncertainty of selecting the most effective model combinations. Our new pretrained pipelines simplify these processes, offering a seamless, user-friendly experience.

A New Augmented NER Model for Multilingual Name Extraction by Leveraging the Capabilities of the LangTest Library to Boost Their Robustness Significantly

The newly introduced augmented NER model namely ner_deid_name_multilingual_clinical_langtest is powered by the innovative LangTest library. This cutting-edge NLP toolkit is at the forefront of language processing advancements, incorporating state-of-the-art techniques and algorithms to enhance the capabilities of our models significantly.

Example:

ner = MedicalNerModel.pretrained("ner_deid_name_multilingual_clinical_langtest", "xx", "clinical/models") \
    .setInputCols(["sentence", "token", "embeddings"]) \
    .setOutputCol("ner")


text_list = [
    """Record date: 2093-01-13, David Hale, M.D., Name: Hendrickson, Ora MR. # 7194334 Date: 01/13/93 PCP: Oliveira, 25 years old, Record date: 1-11-2000. Cocke County Baptist Hospital. 0295 Keats Street. Phone +1 (302) 786-5227. The patient's complaints first surfaced when he started working for Brothers Coal-Mine.""",     
    """J'ai vu en consultation Michel Martinez (49 ans) adressé au Centre Hospitalier De Plaisir pour un diabète mal contrôlé avec des symptômes datant de Mars 2015.""",   
    """Michael Berger wird am Morgen des 12 Dezember 2018 ins St. Elisabeth-Krankenhaus in Bad Kissingen eingeliefert. Herr Berger ist 76 Jahre alt und hat zu viel Wasser in den Beinen."""    ,
    """Ho visto Gastone Montanariello (49 anni) riferito all' Ospedale San Camillo per diabete mal controllato con sintomi risalenti a marzo 2015.""",
    """Antonio Miguel Martínez, un varón de 35 años de edad, de profesión auxiliar de enfermería y nacido en Cadiz, España. Aún no estaba vacunado, se infectó con Covid-19 el dia 14 de Marzo y tuvo que ir al Hospital. Fue tratado con anticuerpos monoclonales en la Clinica San Carlos.""",
    """Detalhes do paciente:
    Nome do paciente: Pedro Gonçalves NHC: 2569870 Endereço: Rua Das Flores 23. Cidade/ Província: Porto Código Postal: 21754-987 Dados de cuidados Data de nascimento: 10/10/1963 Idade: 53 anos Data de admissão: 17/06/2016 Doutora: Maria Santos""",
    """Spitalul Pentru Ochi de Deal, Drumul Oprea Nr. 972 Vaslui, 737405 România Tel: +40(235)413773 Data setului de analize: 25 May 2022 15:36:00 Nume&Prenume: BUREAN MARIA, Varsta: 77 CNP: 2450502264401"""
]

Result:

Please check: ner_deid_name_multilingual_clinical_langtest

DatasetInfo Parameter Added to SentenceEntityResolver Annotator to Track the Source Datasets’ Versions.

Introduced a setDatasetInfo param to SentenceEntityResolverApproach annotator to let users add dataset information (version, year, etc.) to the “model metadata” not the output.

Example:

bertExtractor = SentenceEntityResolverApproach()\
  .setNeighbours(25)\
  .setThreshold(1000)\
  .setInputCols("bert_embeddings")\
  .setNormalizedCol("concept_name")\
  .setLabelCol("conceptId")\
  .setOutputCol('snomed_code')\
  .setDistanceFunction("EUCLIDIAN")\
  .setCaseSensitive(False)\
  .setDatasetInfo("the model version:531")

Robust Exception Handling to Allow Skipping only the Corrupted Records Processed via GenericClassifier, BertSentenceChunkEmbeddings, AssertionFilterer, ChunkFilterer, ContextualParser, ChunkMerge and Deidentification Annotators

We added the doExceptionHandling parameter into GenericClassifier, BertSentenceChunkEmbeddings, AssertionFilterer, ChunkFilterer, ContextualParser, ChunkMerge and Deidentification annotators for a robust exception handling if the process is broken down due to corrupted inputs. Suppose it is set as True. In that case, the annotator tries to process as usual and if exception-causing data (e.g. corrupted record/ document) is passed to the annotator, an exception warning is emitted which has the exception message. Processing continues with the next one while the rest of the records within the same batch are parsed without interruption. The default behavior is False and will throw an exception and break the process to inform users.

Example:

deidentification = DeIdentification() \
    .setInputCols(["sentence", "token", "ner_chunk"]) \
    .setOutputCol("deidentified") \
    .setMode("mask")\
    .setDoExceptionHandling(True)

Changed the license of CPT and MedDRA Models in the ModelHub, and Attempting to Use Them in Healthcare NLP now Throws an Error

The CPT and MedDRA models have been removed from the S3 storage. As a result, when attempting to use these models in Spark NLP, a new error message is thrown. The new error message states that the specified model (e.g., ‘aaa’) cannot be used as a pre-trained model. It further explains that to load the model locally using the .load() method, possession of a valid MedDRA or CPT license is required. If the user has such a license obtained through the corresponding agencies, they are instructed to contact the support team at support@johnsnowlabs.com to inquire about using the model.

Various Core Improvements: Bug Fixes, Enhanced Overall Robustness, And Reliability Of Spark NLP For Healthcare

• Fixed sentence positions in MedicalBertForSequenceClassification
• Updated Deidentification Module according to the latest spark versions
• Updated ALAB Module for assertion result according to tokenization flexibility
• Deprecation of the setRel Method in ChunkMapper: Transitioning to the setRels parameter
• Enhancements in SentenceEntityResolver: Bug Fix and Annotator Refactor
• Added assertion_source, ner_chunk, and ner_label metadata fields to the AssertionDL and AssertionLogReg annotators
• Implemented fixes and enhancements related to entity handling and resolution in Resolver and ChunkMapper, including incorporating an entity field in resolver metadata from embeddings, rectifying the entity field assignment in ChunkMapper, and resolving a bug with all_k_resolutions when using setMultivaluesRelations(True) in ChunkMapper
• Updated init.py and the path of class for BertSentenceChunkEmbedding

Updated Notebooks And Demonstrations For making Spark NLP For Healthcare Easier To Navigate And Understand

• New Task Based Clinical Pretrained Pipelines Notebook
• Updated Clinical Assertion Model Notebook with AssertionMerger example
• Updated Snomed Entity Resolver Model Training Notebook with new parameter
• New Response to Treatment Classification Demo
• Updated Opioid Demo with GPT4 comparison

We Have Added And Updated A Substantial Number Of New Clinical Models And Pipelines, Further Solidifying Our Offering In The Healthcare Domain.

• sbiobertresolve_snomed_no_class
• sbiobertresolve_snomed_conditions
• sbiobertresolve_meddra_lowest_level_term
• sbiobertresolve_meddra_preferred_term
• sbiobertresolve_snomed_bodyStructure
• sbiobertresolve_snomed_drug
• sbiobertresolve_snomed_findings_aux_concepts
• ner_deid_name_multilingual_clinical_langtest
• explain_clinical_doc_ade
• explain_clinical_doc_biomarker
• explain_clinical_doc_public_health
• explain_clinical_doc_risk_factors
• meddra_llt_resolver_pipeline
• meddra_pt_resolver_pipeline
• medication_resolver_pipeline
• medication_resolver_transform_pipeline
• ner_medication_pipeline
• icd10cm_rxnorm_resolver_pipeline
• snomed_term_resolver_pipeline
• snomed_findings_resolver_pipeline
• snomed_body_structure_resolver_pipeline
• snomed_auxConcepts_resolver_pipeline
• snomed_conditions_resolver_pipeline
• snomed_drug_resolver_pipeline
• snomed_resolver_pipeline
• clinical_deidentification_generic_optimized
• clinical_deidentification_nameAugmented
• clinical_deidentification_subentity_optimized
• umls_rxnorm_mapper
• icd10_meddra_llt_mapper
• meddra_llt_icd10_mapper
• icd10_meddra_pt_mapper
• meddra_pt_icd10_mapper
• meddra_llt_pt_mapper
• meddra_pt_llt_mapper
• rxnorm_umls_mapper
• drug_matcher
• ner_opioid
• clinical_deidentification
• clinical_deidentification_multi_mode_output
• bert_sequence_classifier_response_to_treatment

For all Spark NLP for Healthcare models, please check: Models Hub Page

Versions