1 TDT4210 Lecture Sept. 15th 2005: Representing Clinical Information
2 Outline Standardizing clinical information The Unified Medical Language System (UMLS) GALEN Clinical Communication –Clinical Headings –Health Level Seven Reference Information Model (HL7 RIM)
3 Standardization Standardization describes and provides the quality requirements that products, processes, and services must fulfill. It is a systematic activity that creates order, makes selections, and formulates rules. Definition (ISO/IEC): Standardization Activity of establishing, with regard to actual or potential problems, provisions for common and repeated use, aimed at the achievement of the optimum degree of order in a given context.
4 Standards Definition (ISO/IEC): Standard A document, established by consensus and approved by a recognized body that provides, for common and repeated use, rules, guidelines or characteristics for activities, or their results, aimed at the achievement of the optimum degree of order in a given context. The standardization process must be open so that all parties can be represented. Consensus requires that all views and opinions be considered and that concerted action be taken toward their resolution. Consensus means much more than decision by a simple majority, but not necessarily unanimity.
5 Challenges Most terminologies and vocabularies are incompatible A standard terminology will have to serve conflicting purposes due to the multitude of tasks and users Expressiveness vs. tractability
6 Unified Medical Language System (UMLS) R&D project established by NLM (US National Library of Medicine) in 1986 Focus on integration and standardization of clinical vocabularies and terminologies UMLS knowledge sources: –Metathesaurus: Definition of concepts –SPECIALIST Lexicon –Semantic Network: Definition of relations UMLS functions as a reference for concepts and translations of concepts
7 The UMLS Metathesaurus The Metathesaurus contains information about over 1 million biomedical concepts and 5 million concept names from more than 100 controlled vocabularies and classifications (some in multiple languages) used in patient records, administrative health data, bibliographic and full-text databases and expert systems. It includes vocabularies and coding systems designated as U.S. standards for the exchange of administrative and clinical data, including SNOMED CT®SNOMED CT Organized by concept or meaning
8 The UMLS SPECIALIST Lexicon The SPECIALIST lexicon has been developed to provide the lexical information needed for the SPECIALIST Natural Language Processing System. It is intended to be a general English lexicon that includes many biomedical terms. Coverage includes both commonly occurring English words and biomedical vocabulary. The lexicon entry for each word or term records the syntactic, morphological, and orthographic information needed by the SPECIALIST natural language processing system.
9 The UMLS Semantic Network The Network provides a consistent categorization of all concepts represented in the UMLS Metathesaurus®. Each Metathesaurus concept is assigned at least one semantic type from the Semantic Network. Semantic categorization of a wide range of terminology in multiple domains The links between the semantic types provide the structure for the network and represent important relationships in the biomedical domain. The primary link between the semantic types is the 'isa' link. Th
11 GALEN – General Architecture for Languages, Encyclopædias, and Nomenclatures in Medicine EU founded project, consortium of universities and vendors Focus on developing a robust representation formalism to standardize representation of clinical information GRAIL (The GALEN Representation and Integration Language) The GRAIL forms part of the wide range of medical terminology models, methods, architectures and tools developed by GALEN and successor projects. The GRAIL is a concept modeling language for developing computer-based medical terminologies. GRAIL is a compositional and generative semantic network knowledge representation language. The GRAIL representation scheme formed the basis of the GALEN Common Reference Model.
12 GRAIL The GALEN representation and integration language (GRAIL) has been developed to support effective clinical user interfaces and extensible re- usable models of medical terminology. It has been used successfully to develop the prototype GALEN common reference (CORE) model for medical terminology and for a series of projects in clinical user interfaces within the GALEN and PEN&PAD projects. Terminology Server: Delivers terminology services, e.g.: –Concepts –Reference models –Code conversion between thesauri, nomenclature, classifications, local controlled vocabularies, formal models, and multi-lingual support. Strengths: –Uniform representation of medical concepts –Supports a multitude of tasks
13 The GRAIL language
14 GALEN Terminology Server Architecture discriminating strictly among concepts, languages, coding systems, and general inference modules. Produced a unified controlled medical vocabulary through four components; –A CORE model –A GRAIL formalism –A code-conversion module –A multi-lingual module
15 GALEN Terminology Server
16 Clinical Communication Clinicians often use clinical terms different from each other, or with different meanings Two concepts that define and reference clinical information: –National Health Services (NHS) Clinical Headings Project –Health Level 7 RIM
17 Clinical Headings Main objective: To identify a limited number of headings that clinicians need for communicating clinical information Headings function as a tool for common reference and navigation in clinical information Clinical headings refer to expressions and phrases that clinicians use to name sections of their communications (e.g. symptoms, examinations, test results, diagnosis, etc.)
18 Clinical Headings
19 Health Level Seven (HL-7) (http://www.hl7.org/) ANSI accredited standards development organization with close liaison to ISO TC 215 Provides a standard for interchange of clinical data Specifies communication contents and exchange formats on the application layer (ISO/OSI 7th layer) Model-based specification of messages on the basis of a Reference Information Model (RIM) –Partially object-oriented methodology –Not a model for database storage
20 A domain-specific, common protocol for the exchange of health care information. 1Physical 2Data Link 3Network 4Transport Communication 5Session 6Presentation 7Application Function What does “HL7” stand for? ISO-OSI Communication Architecture Model From tutorial by George W. Beeler, Dresden, August 2000
21 HL7 RIM Used to express the information content for the collective work of the HL7 Working Group. A coherent, shared information model that is the source for the data content of all HL7 messages. Maintained by a collaborative, consensus building process involving all Technical Committees and Special Interest Groups. RIM change proposals are debated, enhanced, and reconciled by technical committee representatives and applied to the RIM during the model harmonization process From tutorial by George W. Beeler, Dresden, August 2000
22 RIM Domain Subject Areas Stakeholders –Patient –Person –Stakeholder –Healthcare_service_ provider –Organization Healthcare finances –Patient_billing_account –Healthcare_benefit_plan –Guarantor_contract Patient_encounters –Clinical Documents –Scheduling –Patient_service_material –Patient_encounter –Patient_service_event Master_tables Message_control From tutorial by George W. Beeler, Dresden, August 2000
23 Initial taxonomy of classes Subject classes are the focus of a committee’s interest. State- transition models are defined for subject classes. Messages are defined based on a state-transition model for subject classes. Associative classes describe complex logical association between other classes. Roles, events, and participation classes are special kinds of associative classes. Substantial classes represent some ‘obvious’ concept of the application domain. –are frequently subject classes –are usually not merely associative classes –should have an identifier attribute From tutorial by George W. Beeler, Dresden, August 2000
24 RIM core classes Entities –Physical things or organizations, or groupings of physical things; e.g. organisation, living subjects, materials, location) Roles –Played by those entities and therefore assigning them the competence to perform specific actions, e.g. patient, provider, employee, specimen, practitioner) Participations –Of role playing entities in specific acts, e.g. performer, author, subject, destination, witness) Acts –An instance of Act represents a record of an intentional action in health care; e.g. observation, procedure, supply, medication Role Relationships –to manage interactions between entities in their corresponding roles Act Releactionsip –chaining different acts. From tutorial by George W. Beeler, Dresden, August 2000
25 RIM Class Diagram - v0.96 From tutorial by George W. Beeler, Dresden, August 2000
26 Data Types Patient Encounters USAMMessage Stakeholders Material Finance Scheduling Document Stakeholders Stakeholder Affiliations Data types Service Provider AppointmentsEncounterEpisode USAMServiceMessagecontrolDoc.Roles Legend Person Patient Material Organizations Location Material Material Insurance Accounting Finance RIM Class Diagram - v0.96
27. Patient Encounters SchedulingDocument Stakeholders Stakeholder Affiliations Data types Service Provider Appointments EncounterEpisode USAMServiceMessagecontrol Doc.RolesLegend Person Patient Material Organizations Location Material Material Insurance Accounting Finance RIM Class Diagram - v0.96 From tutorial by George W. Beeler, Dresden, August 2000
28 Current 0.96 RIM Statistics 110 Classes 532 Attributes 167 Association Relationships 30 Inheritance Relationships 2 Aggregation Relationships 37 Subject Areas –18 Domain, 8 Work-in-Progress, 12 Administrative 43 Data types Maintained in an Access database, expressed in UML, and annotated with a literary expression From tutorial by George W. Beeler, Dresden, August 2000
29 Literature/links: van Bemmel & Musen: Handbook of Medical Informatics, ch. 29 and 34 thesis.pdf, ch. 4http://www.idi.ntnu.no/grupper/su/publ/phd/bayegan- thesis.pdf model/V3_Tutorials/tut2_index.htmhttp://www.hl7.org/library/data- model/V3_Tutorials/tut2_index.htm
30 Next week Knowledge Representation (Thomas) Introduction of exercise no. 1