What s next for openehr Sam Heard Thomas Beale
Current situation (2010-) General industry movement toward SOA, growing use of terminology SOA: IHE Information / messages HL7 v3 failed, org in fresh look EN13606 limited uptake Terminology IHTSDO: SNOMED CT, ICD11 HL7: LOINC DCMs
SOA & IHE Due to negative experiences with overcomplex data standards, some parts of industry moving to far simplified data, more communication e.g. XDS / CDA services. This addresses some short-term needs Provides needed services e.g. PIX, PDQ However, these don t solve the long-term need, which is computable data.
EN13606 openehr developers active in CEN for 4 years 2001-2005 openehr archetype model (ADL 1.4) used as EN13606-2 EN13606-1 heavily based on openehr circa 2006
EN13606 - status Industry uptake appears limited Some countries like Sweden nominally mandated it, but not yet implementable Limitations: Not yet proven in any large scale system Fixed demographic model (no archetyping) EHR Extract model not very flexible e.g. For multiple patients Uncertainty over ISO 21090 standard
EN13606 - status Coming up for 3y review in ISO May 2012 If demonstrable industry uptake revision Else, obsolete openehr proposal: Part 1 and openehr RM merged Part 2 be upgraded to openehr ADL/AOM 1.5
EN13606 & openehr Extract openehr Extract far more flexible, subsumes 13606 data Demonstrated in Singapore, Sweden ADL 1.5 archetypes available online Will be offered to 2012 ISO review process
HL7 HL7v3 messaging has not been used widely, not performed well where it was used. Now appears limited to CA, NL, with legacy in UK CDA is in fairly common use in some countries Essentially an RMIM also an XSD 3 levels level 3 = structured data So far, limited use for structured data HL7v3 RIM basis has meant still too complex for many green CDA
HL7 fresh look HL7 has recognised the limitations of the v3 effort and low take-up Efforts underway to make CDA more computable, easier to use However, still not model-based XSD only Grahame Grieve (Australia) has proposed complete v3 replacement called Resources for Health (RFH) New data types Re-engineered data models
HL7v3 Replacement proposal
openehr and HL7v2 Integration with HL7v2 tractable and proven: openehr Template per message type + TDS Can use tools like BizTalk to process large amounts of data Standardised mappings to v2 message types may appear on openehr.org in the future Commonly undertaken in real systems today via TDS
openehr and HL7 CDA General approach: Base schema templates = restricted schemas Problems: Templates not re-usable Entry types based on RIM, and hard to manage HL7 data types problems Green CDA may improve things
openehr and CDA Integration is possible, but requires a more complex framework to deal with structured level 3 content Has been demonstrated with Australian discharge summary
Detailed Clinical Models (DCMs) ISO work item on quality criteria for models In very early stages Confuses need for representation with TF view: nothing should be standardised without prior industry experience International group led by Stan Huff (IHC) now meeting to choose a common formalism
News flash - CIMI forum International forum to decide shared clinical model format Four Meetings between May 2011, Nov 2011 Process: initial vote for 2 out of 5 candidates ADL 1.4, ADL 1.5, HL7 RIM, OWL, UML (various forms) voted on 20 Nov ADL 1.5, UML
International CIMI forum Run by Stan Huff, Intermountain Healthcare 1. Establish and maintain a shared repository of DCMs where the models are: described using a single formalism built using an agreed upon set of base data types with agreed behaviors (constraints, ordering, defaults, handling exceptions, runtime processing assumptions, etc.) the formalism must support the capability to formally bind to terminologies for both data element names and for values of data elements strategy for defining terminology concepts when they do not exist that does not stop the creation or use of the models where the repository is open and contents are free for use at no cost, to enable use in healthcare systems globally and is also open for contributions, to allow contributors to work on content that they see as priority use cases and maintain provenance for the models including examples of instances (over time) for data for the models expressed in a well defined reference syntax
International CIMI forum 2. Establish appropriate governance, finance, and organizational structure for the group 3. Capture background, context, and discussions relevant to decisions made in creating and using the model Notes on use, problems, errors, and misuse 4. Establish a fair and open process for curating the content in the model repository Adherence to the single approved formalism Verification by clinicians and other domain experts Endorsements (approvals) by specific organizations or disciplines Checking for consistency to prevent duplicate and overlapping content Manage dependencies and life cycle
International CIMI forum 5. Promote the creation of software tools that translate the models to other commonly used model representation languages (OWL, UML, HL7 MIF, graphic format, HTML, etc.) 6. Promote the creation of software tools that generate implementation artifacts (XML schemas, Java class definitions, CDA templates, GreenCDA, etc.)
CIMI forum members Providers Intermountain Health, Utah Mayo Clinic US Department of Defence Kaiser Permanente US Veterans Health Administration (VHA) Vendors GE Harvard / Smarthealth National programmes UK NHS, Singapore MOH, Canada Infoway, Nehta, Sweden Organisations openehr IHTSDO HL7
International CIMI forum Outcome of London meeting openehr ADL / AOM 1.5 chosen as starting point formalism for internationally shareable clinical models UML 2.0 profile will be created based on openehr AOM 1.5 This will connect major UML-based tooling efforts to openehr semantics all shared models will be archetypes Bridges to generate HL7 clinical statement pattern and other concrete formats
IHTSDO Board members are stakeholders nat.gov programmes Structure international group + National Release Centres (NRCs) Structure of SNOMED CT mimics international + NRCs int l release + national extensions Ref set model specified most important for next 5y DCM binding a key interest area
openehr & IHTSDO 2009 IHTSDO & openehr announced concept of merger; backed by IHTSDO board 2010 members forum voted against (US, CA) 2011 IHTSDO has indicated that a formal relationship is still under consideration Archetype / snomed bindings
openehr Governance under review More open specification development Software development like Apache Clinical modelling CIMI activity has brought much wider interest Archetype technology will now be of central international importance going forward
openehr technical directions openehr 2.x reference model underway Service specifications & IHE mappings being worked on openehr RM + archetypes will be used to update 13606 for 2012 13606 revision RM and part 2 (archetypes) Process level models (care pathways and active indexes) will be published TDS / TDO specifications will be published Big increase in software development
openehr clinical directions More involvement in real clinical trials Expand archetype editor team Higher rate of archetype publishing Establish Clinical Knowledge Incubator Open source tool Informal early archetype development? Regional instances
openehr - organisation More focus on localisation activities Help for national chapters to be set up current: Brazil, NZ, Japan, Chile Aiming to have 1 conference per year first might be in Lake Bled
Where in use? Industry GOV Academia