Keywords: RDF, OWL, Ontologies, e-Health, ebXML Registry Repository, BCM Business Centric Methodology, Templates, Archetypes, Semantic Web, HL7, EHR
Biography
Carl Mattocks is Founder / CEO of CHECKMi a New Jersey-based company developing agent –based software capable of linking a 1000 databases into a single compendium. Commencing with the ISO IRDS standard for Repositories - Carl is a long time contributor to the development and exploitation of metadata-centric specifications. Currently he acts as a co-chair for the Business Centric Methodology TC and the ebXML Registry Semantic Content SC. Much of Carl’s wisdom was gained managing a London–based Artificial Intelligence company that pioneered the use of faceted-classification techniques and multi-national language processing for online information discovery. He has also benefited from being - one of the first Certified Systems Analysts in the UK; one of the first students in a condensed-MBA program at Cranfield School of Management; and one of the first designers to apply (ITIL) Structured System techniques on (Government / Financial Service) intelligence projects.
This paper outlines how the knowledge embedded in OWL and medical ontologies support e-Health web service deployment that may involve the interaction of independent resources from different domains and owned by different entities. In particular, it (1) identifies how the OASIS Business Centric Methodology guidelines apply to the development and publishing challenges for medical information that is required by agents representing the e-Health stakeholders and (2) how the enterprise content management capabilities of the ISO/TS 15000 ebXML Registry / Repository are being used to manage standards based (e.g., HL7 CDA, GEHR) medical records and assist the discovery process.
1. Introduction
2. e-Health Service
3. Business Centric Guidelines
4. E-Business Registry
5. Semantic Web
6. Medical Documents
6.1 CEN ENV 13606-2
6.1.1 Good European Health Record
6.1.2 openEHR
6.1.3 Clinical Document Architecture
7. Clinical Information
8. Ontology Management
9. Choice Points
10. Semantic To Do
Acknowledgements
“ e-Health refers to the use of modern information and communication technologies to meet needs of citizens, patients, healthcare professionals, healthcare providers, as well as policy makers .” - Declaration made during the European Commission 2003 e-Health Ministerial Conference
The Semantic Web is about the autonomous discovery and assembly of distributed remote resources. e-Health services depend on the automated sharing of metadata across web applications that provide a common approach for the discovery, understanding, and exchange of semantically rich Electronic Health Record (EHR) information. This paper outlines how electronic medical records information embedded in an e-business Registry & Repository support e-health service goals that involve the interaction of independent resources from different domains and owned by different entities. In particular, it identifies Business Centric Methodology (BCM) guidelines for the development and publishing of the semantic content required by agents representing the Health Management stakeholders that use OWL, ontologies, templates and medical information archetypes to both assist the discovery process and pull health care services together.
Service Goals
Across the world many governments are announcing e-Health service initiatives. For example, in July 2004 the USA Health & Human Services (HHS) Secretary outlined a 10-year plan to transform the delivery of health care by using electronic health records and accelerate regulations for e-prescribing drugs. Within the OASIS open source specifications body there are a number of Technical Committee (TC) groups actively contributing to the evolution of e-Health service oriented standards. Specifically, this paper references the work of TCs for (i) the Business-Centric Methodology (BCM), and (ii) the ebXMLRegistry to help explain how the application of standards based technology support the e-Health goals of:
E-Prescribing
In the USA there are more than 3 billion prescriptions written annually and the national savings from universal adoption could be as high as $27 billion. Savings that derive from providing instant electronic connectivity between the practice, the pharmacy, health plans / PBM’s (pharmacy benefit management), and other agencies, see Figure 1. Specifically, a key part of the USA HHS plan is that, internet technology will allow beneficiaries to access their personal health care information and that electronic prescribing will improve quality, efficiency, and reduce cost by the facilitation of:
Simple electronic prescribing systems include electronic communications between the electronic prescribing system and the clinician and pharmacy (fully electronic or via fax). Alternative communications are still required with the health plan and the patient, and usually occur by phone (dashed lines). In a fully integrated system, all of these communications can be done electronicallyElectronic Prescribing: Toward Maximum Value and Rapid AdoptionElectronic Prescribing Initiative
Forms & Vocabulary
An eHealth Initiative report that identified the benefits of electronic connectivity also stated that a number of enhancements in semantics are needed to improve quality, efficiency, and to facilitate interoperability between the various electronic systems especially those involved in the electronic prescribing process. In particular it noted that, unifying state (electronic) prescription-form standards, establishing a consistent “doctor-level” drug vocabulary, and standardizing formulary information are among the highest needs. As a guideline to web-based interoperability the OASIS BCM (Business-Centric Methodology) states that a proper interpretation of the business language semantics found in an e-Health SOA (Service Oriented Architecture) is essential for harnessing tacit knowledge and facilitating shared communications. Particularly, the BCM identifies that a Conceptual Layer must enable the exploitation of community-of-interest specific ontologies (as in, EHR vocabularies and code lists supporting electronic form / message validation) is a key factor in semantic interoperability. Further, the ontology capability must be rich enough to resolve all semantic (meaning & operability) conflicts over terminology used to populate the many building blocks of the Lubash Pyramid, see Figure 2.
Ontology capability spans the building blocks of the BCM Lubash Pyramid
Figure 2
While not defining a mandatory structure, BCM states that the Conceptual Layer consists of semantic relationships and controlled vocabularies that increase the meaning of schema (e.g. electronic form) metadata and provide contextual validation to items that have metadata properties (e.g. archetypes and templates). The simplest form of this being a data dictionary that contains metadata about data elements and their relationship between simple / complex data types and their valid value ranges aka archetypes. BCM expects that when recorded in a registry the Conceptual Layer has the role of:
Federated Content
The BCM also identifies that a registry combined with a repository is a key factor in the management of service-oriented components. Such as, metadata registered about electronic form / electronic messaging schemas, schema archetypes (the allowable values for the data type), associations between elements and any stored artifacts. Wherein, a registry not only acts as an interface to a repository of stored content, it formalizes how information is to be registered and shared. Since, in e-Health Services this sharing goes beyond a single enterprise or agency, this dictates that the registry catalog must be capable of supporting metadata used for federated content management. Noting that, a federated content management capability is required when there is as a need for securely managing and accessing metadata across physical boundaries. Irrespective of the boundary type (e.g. community-of-interest, system, department, or enterprise separation), federated content management enables information users to seamlessly access, share and perform analysis on information. For e-Health services this may include:
ebXML Registry
ebXML (e-business XML) is a suite of [ISO/TS 15000] standards from the e-business community addressing the entire Service Oriented Architecture lifecycle of business to business and business to consumer activities, such as, e-Health Services. In particular, when following BCM (Business-Centric Methodology) guidelines, SOA / ebXML service description and discovery extends beyond classification and resource location towards the support of web service activities, such as, e-prescribing. Specifically, when a service is profiled within a federated ebXML registry , the meta-information identifies both the SOA business conditions and the policy rules of access per level of privilege. For instance, a successful e-prescription electronic communications that passes between USA based clinician, pharmacy service and payment provider will need to employ some form of contractual electronic exchange referencing drug labeling and drug listing information maintained by the Food and Drug Administration (FDA) and the National Library of Medicine (NLM)
Fortunately, to manage this knowledge, e-Health services can exploit the OASIS ebXML Business Process ( ebXML BP ) standard that defines metadata describing the service capabilities needed to support electronic business collaborations with default error state resolution. For example, an “e-contract” employing terms of workflow can be structured using Collaboration Protocol Profiles (CPPs) and Collaboration Protocol Agreements (CPAs). ebXML BP also defines and describes activities during service enactment for the transactions (including point-in-time parameters) that have to be fulfilled according to the services’ usage terms.
Registry & Repository
A foundation of the [ISO/TS 15000] suite is the EbXML Registry which specifies the components for web based registry and repository services. In terms of publishing content the ebXML Registry / Repository specification supports:
Its federation ability has won it many supporters including the UN/CEFACT Information Content Management Group (ICG) which has officially adopted the ebXML Registry standard and the open source code freebXML Registry as the center piece of their new Information Content Management Architecture. Version 3 of the ebXML Registry / Repository supports the following types of cooperating registry services:
Access to Artefacts and Artifacts
Acknowledging that there are therefore two basic models of distributed information - a central repository of shared items (with individual entitiess uploading and downloading as required) or a fully distributed model (with the repository distributed over multiple facilities) the ebXML Registry specification supports a single access to many federated Registry / Repository facilities. Thus, it allows:
Equally, to ease discovery and deployment of the collective artifacts ( aka UK artefacts) the ebXML Registry RIM (Repository Information Model) explicitly supports many Classification Schemes. For instance, ebXML Registry enables one or more of the objects defining an e-Health Web Service content ( a “Service” class RegistryEntry) to be classified using Taxonomy Values ( ClassificationNode within a ClassificationScheme). Additionally, a WSDL (Web Services Description Language ) description of the service instance (i.e. as a technical specification file) may be stored as ExtrinsicObjects. Wherein, the relationship between the description files and the “Service” class is established through the “ServiceBinding” Class of ebXML. See Figure 3
ebXML Registry Classification subset of Reference Information Model
Figure 3
WSDL Note: WSDL is an XML format for describing network services as a set of endpoints operating on messages containing either document-oriented or procedure-oriented information. The operations and messages are described abstractly, and then bound to a concrete network protocol and message format to define an endpoint. Related concrete endpoints are combined into abstract endpoints (services). When managing WSDL and XSD Schemas the classification scheme provides for a number of uses:
WC3 RDF
The World Wide Web was originally built for human consumption, and although everything on it is machine-readable , this data is not machine-understandable . The Resource Description Framework (RDF) is a WC3 standard based on the idea of identifying things using Web identifiers (URIs), and describing resources in terms of simple properties and property values. This enables RDF to represent simple statements about resources as a graph of nodes and arcs representing the resources, and their properties and values. For example, the group of statements "there is a Person identified by http://www.w3.org/People/EM/contact#me, whose name is Eric Miller, whose email address is em@w3.org, and whose title is Dr." could be represented as the RDF graph below Figure 4. Adding a layer of semantic information to that defined in XSD schemas RDF also provides an XML-based syntax (called) for recording and exchanging these graphs.
<?xml version="1.0"?>
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:contact="http://www.w3.org/2000/10/swap/pim/contact#">
<contact:Person rdf:about="http://www.w3.org/People/EM/contact#me">
<contact:fullName>Eric Miller</contact:fullName>
<contact:mailbox rdf:resource="mailto:em@w3.org"/>
<contact:personalTitle>Dr.</contact:personalTitle>
</contact:Person>
</rdf:RDF>
Plus OWL provides support for axioms enables assertions of subsumption or equivalence with respect to RDF / OWL classes or properties as in:
OWL has three increasingly-expressive sublanguages: OWL Lite, OWL DL, and OWL Full. The OWL-based Web Service Ontology (OWL-S) Version 1.0 "features a number of refinements to the Service Profile and Process Model. A method of registering (in UDDI & EbXMLRegistry) and linking descriptors of Web service, WSDL and OWL-S has been outlined by researchers in Turkey. A key factor is the Service Profile which is used to concisely represent the service in terms of capabilities, provenance, and operational parameters ( e.g. cost-of-use, quality-of-service parameters, etc), for constructing both advertisements and requests, such as:
Health Standards Vocabularies
RDFS & OWL semantics are domain dependent; thus to be useful domain knowledge is required. E-Health service is one of the few domains to have extensive domain knowledge exposed through standards for describing electronic healthcare records. For example, Health Level 7 (HL7) is health information standards community that has its V2 messaging standards are in wide use around the world mainly for application-level messaging and clinical document management.
The HL7 Version 3 standard is based upon a Reference Information Model (RIM) that abstractly describes messages medical events and transactions. Message content schemas are derived by a restriction process which starts from the Reference Information Model (RIM), and continues through domain information models (DIMs), restricted message information models (RMIMs), common message element types (CMETs), finally ending with hierarchical message definitions (HMDs) and generated message schemas in XML. For instance, an e-prescription would reference multiple DMIM's (Domain Message Information Model) such as Orders and Observations, Pharmacy, Medications, Patient Administration (for patient and clinician identifiers) and diagnostic indications. A feature of the Version 3 methodology is the specification of vocabularies or “value sets” that convey the payload of a specific message e.g. an e-prescription message defining the ordered drug, form, dose, route, and patient instructions.
CEN ENV 13606-2 / GEHR/ openEHR / Clinical Document Architecture (CDA)
Beyond HL7, e-Health services can elect to leverage the CEN ENV 13606-2 information models which provide a means to represent the original organizational structure of one or more nested electronic healthcare record entries. This standard proposes sub-categorization of the EHR into four specializations:
In addition, the Good European Health Record, was produced by a European Health Telematics research programme that developed a comprehensive multi-media data architecture for using and sharing electronic healthcare records, meeting clinical, technical, educational and ethico-legal requirements. Later (mostly australian) extensions on the GEHR approach uses a formal semantic model, known as the GEHR Object Model (GOM). Whereas, rather than try to model all possible clinical concepts, the GOM provides concepts at a number of levels:
For example, clinical models are expressed outside the GOM in the form of XML-Schema archetypes . These archetypes act as constraint definitions and define how to create clinically valid structures out of the GOM primitives. Based on Australian researchers experience using XML based metadata it was found that archetypes are needed for:
More recently, the open EHR specifications incorporate a two-model approach that allows the separation of generic features of any EHR from the domain-specific (usually clinical) features needed for specific EHR instances. Specifically, open EHR health record consists of 'container classes' which contain information about the patient or data subject :
Composition (or document) - this is the class that contains information committed to the EHR by a clinician.
Section - this class allows information within a composition to be segmented
Entry - this class contains meaningful information that is to be processed by the machine and read by the clinician.
Note : These classes contain no clinical or demographic concepts at all - and it is this feature which differentiates the open EHR approach. The clinical (model) requirements are met through archetypes and templates designed for the purpose.
The HL7 Clinical Document Architecture ( CDA ) is a generic model for the communication of clinical documents, very similar to the "Composition" class in the CEN 13606 specification, the "Transaction" class in openEHR and the HL7 record architecture. CDA documents are encoded in Extensible Markup Language (XML).
Examples of LOINC a set of more than 10,000 names and codes developed for use as observation identifiers in standardized messages exchanged between clinical computer systemsLOINC_NUM COMPONENT (Type of Service) SYSTEM (Setting) METHOD_TYPE (Subject Matter Domain and/or Training / Professional Level)
34128-9
SUBSEQUENT EVALUATION NOTE
OUTPATIENT
DENTISTRY
34901-9
SUBSEQUENT EVALUATION NOTE
OUTPATIENT
GENERAL MEDICINE
34132-1
SUBSEQUENT EVALUATION NOTE
OUTPATIENT
PHARMACY
Cross Enterprise Clinical Documents Sharing (XDS)
Integrating the Healthcare Enterprise (IHE) is an initiative designed to stimulate the integration of the information systems that support modern healthcare institutions. It has numerous sponsors and supporting organizations in different medical specialty domains and geographical regions. The IHE Integration Profile known as Cross Enterprise Clinical Documents Sharing is document-content neutral - it will support any type of document without regard to content and format. XDS is focused on providing a standards-based specification for managing the sharing of documents that healthcare entities have decided to explicitly share, such as documents containing simple text, formatted text, images or structured and vocabulary coded clinical information.
Based on the ebXML Registry specification (and implemented with the freebxml registry ) the XDS defines the (document) Registry as an actor that maintains metadata about each registered document in a document entry. It also enforces some healthcare specific technical policies at the time of document registration The registry metadata includes a link to the (document) Repository where the actual document is stored which in turn assigns and maintains a unique identifier for each document, to allow Document Consumers to retrieve them. Some of the key XDS concepts are:
NIST HL7 Experimental Registry
The USA National Institute of Standards and Technology (NIST) HL7 Experimental Registry tool (also using the freebXML Registry is part of a collaborative effort to determine the appropriate environment, policies and procedures for the development, submission, storage and retrieval of HL7 artifacts. Each artifact has Associations that link it to the organizations that submitted it or are responsible for its maintenance. If an artifact references one or more CMETs in its specification, then there is a Uses association to identify the link from that artifact to each of the referenced CMETs. Further, if an artifact specification is a refinement of a vocabulary domain over any of the structural attributes of its classes, then that artifact is classified by the node in the domain hierarchy to which it is constrained. It is NIST’s intent to always have available the most recent DMIMs, RMIMs, HMDs, and MessageTypes from each of the HL7 technical development committees, including:
Mapping & Annotating Ontologies
As required, e-Health service component developers can employ the CEN ENV 13606-2 / GEHR / openEHR / Clinical Document Architecture (CDA) as standards for exchanging information. Wherein, e-Health service information can be managed by two different e-Health service entities using different message structures. To help automate that interoperation the members of the ARTEMIS project are providing a standard way of accessing the data by registering & storing (1) ontologies based on existing healthcare standards and (2) the semantic mapping between these ontologies. Given, when the e-Health (web) services are annotated with these ontologies, it becomes possible for healthcare organizations conforming to different standards to invoke each others web services by semantic mediation.
Further, to be able to discover the services stored in a Registry, the ARTEMIS project leaders have identified the need for semantic service registry query mechanisms that leverage previous research linking OWL to the Registry Information Model objects, see Figure 5.
Figure 1:
Source : Enhancing ebXML Registries to Make them OWL Aware
Leveling Ontologies
A Community of Interest ontology is a key concept of the BCM Conceptual Layer that supports the semantic links between the building blocks that a service employs. This concept follows from the recognition (validated by formal network theory) that interacting entities, whether they be people or enterprises and their technical systems, tend to coalesce in groups with common characteristics, such as purpose, vocabulary and behavior. As recognized by the XDS Affinity Domain those communities may be defined by formal or informal organizational structures. Indeed, since one e-Health service community may be composed of many sub-communities it is recognized that there is a single Community of Interest ontology to support multiple-levels of ontologies. Wherein the scope of:
XForm & Templates
Formally, templates are considered a gauge, pattern, or molded object. The BCM guideline consider templates as a framing mechanism capable of (1) framing the details of a business process and (2) referencing the appropriate domain ontologies and other semantic information defined in the Conceptual Layer . To define their role as an Electronic PRocess (EPR) portal technology a BCM committee is collaborating with Norwegian researchers to specify how semantically rich XML templates, stored in a registered Folder , can support formal interaction among interacting communities such as e-Health, e-Building and e-Government.
The BCM e-Health effort also expects to leverage the emerging HL7 approach of employing templates as a form of constraint statement model, which is directly usable for:
Note: the following template types have been identified by researchers at the Mayo Clinic and other HL7 SIG members
Archetypes & Templates
It is widely acknowledged that the operation of an e-Heath service will involve thousands of variants of business processes, business rules, business patterns, and data permutations. The BCM guidelines refer to the decisions that use these e-business factors as Choice Points. Further, BCM practitioners assert that the explicit identification and management of these Choice Points significantly aids to comprehensibility, alignment, while promoting tracing and accountability. A prime application of the choice point approach are e-Health templates that use archetypes, wherein:
EbXML Registry Semantic Content Management
To ensure that all types of e-Health Service componentry can be registered and stored in a readily accessible manner the members of the OASIS ebXML Registry are currently defining how Semantic Content Management will be facilitated in the next version of the specification. In addition to providing support for semantic links based on RDF and OWL the check list of requirements includes :
XHTML rendition made possible by SchemaSoft's Document Interpreter™ technology.