book

Hands-On Healthcare Data

by Andrew Nguyen

August 2022

Beginner to intermediate

242 pages

7h 22m

English

O'Reilly Media, Inc.

Book available

Read now

Unlock full access

Conventions Used in This BookUsing Code ExamplesO’Reilly Online LearningHow to Contact UsAcknowledgments
The Enterprise MindsetThe Complexity of Healthcare DataSources of Healthcare DataElectronic Health RecordsClaims DataClinical/Disease RegistriesClinical Trials DataData Collection and How That Affects Data ScientistsProspective studiesRetrospective studiesConclusion
Basic Introduction to Docker and ContainersInstalling and Testing DockerConceptual Introduction to DatabasesACID ComplianceOLTP SystemsOLAP SystemsSQL Versus NoSQLSQL Databases(Labeled) Property Graph DatabasesHypergraph DatabasesResource Description Framework DatabasesConclusion
Controlled Vocabularies, Terminologies, and OntologiesKey ConsiderationsPre-coordination Versus Post-coordinationCase Study Example: EHR DataCommon TerminologiesCPTICD-9 and ICD-10LOINCRxNormSNOMED CTKey TakeawaysUsing the Unified Medical Language SystemSome Basic DefinitionsConcept OrientationWorking with the UMLSUMLS and Relational DatabasesPreprocessing the UMLSUMLS and Property Graph DatabasesUMLS and Hypergraph DatabasesReview of the UMLSConclusion
Publicly Accessible DataMedical Information Mart for Intensive CareSyntheaData ModelsGoalsExamples of Data ModelsCase Study: MedicationsThe Medication Harmonization ProblemTechnical Deep DiveConnecting to the UMLSDifficulties Normalizing Structured Medical DataConclusion
Publicly Accessible Data—SynPUFData ModelsChoosing a Data ModelCombining Claims and EHR DataCase Study: Combining Diagnoses and MedicationsOMOP Versus GraphsConsiderations When Combining Different Sources of Healthcare DataConclusion
A Primer on Machine LearningWhat Is Feature Engineering?Graph-Based Deep LearningExtracting Data as a TableTo SQL or Not to SQLQuerying OMOP DataFrom Graphs to DataframesWhy Add the Complexity of Graphs?Machine Learning and Feature Engineering with GraphsGraph Embeddingsnode2veccui2vecmed2vecsnomed2vecSome Final Thoughts About EmbeddingsMaking the Case for Graph-Based AnalysisConclusion
Federated Learning and Federated AnalyticsHow Does Federated Learning Work?Why Federated Analytics/Learning?The Data Harmonization Challenge in a Federated ContextGraphs and Federated ApproachesNatural Language ProcessingConcept ExtractionBeyond Concept ExtractionClinical NLP ToolsCommercial Clinical NLP SolutionsKey Differences Between Clinical NLP and Other Applications of NLPConclusion
Other Types of Healthcare RWDData Normalization and HarmonizationMerging DatasetsBridging IT and the BusinessIt’s a Human, Not Technical, ProblemGraphs Can Be Part of the SolutionGraphs Are Not a Silver BulletConclusion

Content preview from Hands-On Healthcare Data

Chapter 3. Standardized Vocabularies in Healthcare

In this chapter, we cover standardized vocabularies, controlled vocabularies, terminologies, and ontologies. Some use these terms quite interchangeably, while others adhere to fairly precise definitions that may be confusing at first. We will start with a short introduction to these ideas, how they are similar, where they overlap, and also how they differ. You will see me use the term terminology as a nebulous reference to a particular source of content that spans the spectrum of vocabulary to ontology. For example, International Classification of Diseases, Revision 10 (ICD-10), ICD-9, and Systematized Nomenclature of Medicine—Clinical Terms (SNOMED-CT) would all be referred to as terminologies. Some are simply a collection of terms and definitions (e.g., vocabularies), while others capture hierarchies or even richer relationships (e.g., ontologies).

In addition to the content, there are also associated tools and processes that build on top of the semantic web. For example, many ontologies come in Web Ontology Language (OWL) and build on top of Resource Description Framework (RDF). These technologies are not specific to healthcare or medicine, but this is one area where we have seen a larger adoption of these technologies.

If there is a single technical chapter in this book that differentiates it from books about data (especially in other industries), it would be this one. Many of the challenges when working with healthcare data ...