The crisis in ‘standards’ in e-health Thomas Beale September 2009.
-
Upload
luke-jonathan-singleton -
Category
Documents
-
view
214 -
download
2
Transcript of The crisis in ‘standards’ in e-health Thomas Beale September 2009.
The crisis in ‘standards’ in e-health
Thomas BealeSeptember 2009
© Thomas Beale 2009
The Industry Need
Healthcare is an information-intensive business.
Healthcare data is captured piecemeal during clinical work processes but used in different ways by other processes.
Clinical care of patients is shared among multiple provider enterprises (exacerbated by increasingly mobile citizens), requiring information sharing.
© Thomas Beale 2009
The Industry Need
Healthcare is an information-intensive business.
Information needs to be aggregated per-patient to be computable – to allow personalised healthcare & decision support and then across populations, for public health analysis and medical research.
Healthcare information is complex and changing and therefore challenging to manage over time.
© Thomas Beale 2009
Semantic interoperability :==
Meaning preservation – from the user interface to the back-end.
Sharing – getting the data together.
Aggregation – merging data from different sources and computing on it.
Evolution – preserving and adding meaning over time.
© Thomas Beale 2009
What kind of solution do we need?
Is it a message standard? Might be useful…. But what messages? We need hundreds! Ok… in that case, we need some kind of MODEL
or LANGUAGE for expressing messages Maybe we can standardise on that… and define
a method for creating individual message specifications…
That would take care of the data moving between systems…
© Thomas Beale 2009
What kind of solution do we need?
Is it a document standard? If a document is something we want to store in a
system, then such a standard might be useful… But what kinds of documents? There are
thousands of possibilities! Perhaps we also need some kind of MODEL or
LANGUAGE for expressing documents… and a method for defining different types of
documents and variations… That would take care of some data in systems…
© Thomas Beale 2009
What kind of solution do we need?
Is it a SOA standard? Hm….if we could standardise on the
INTERFACEs exposed by data repositories, then anyone could write software to talk to them!
But interfaces contain functions that return data in the form of documents, or messages….
So we need those document and message standards…
© Thomas Beale 2009
What kind of solution do we need?
Is it a User Interface standard? If we want to make sure human users interact
consistently with the software – especially the WORKFLOW, we had better do something about screen layout, widgets etc
But these need to be connected in a known way to the underlying data
Do we need a MODEL of the user interface layout and workflow?
© Thomas Beale 2009
What kind of solution do we need?
Is it a standard terminology? Well, we have standardised terms for some
things, like diseases, lab concepts…. Agreeing to use these would help wouldn’t it? Yes, but they don’t do anything on their own, they
need to be included in documents, messages…and we nearly forgot – the user interface!
© Thomas Beale 2009
What kind of solution do we need?
Is it ontologies? That’s a deep question. What is an ontology?
A: some kind of descriptive model of reality. We usually distinguish ‘upper level ontologies’ which are generic, and domain-specific ontologies
We can think of ontologies as ‘statements of truth’ about things in the domain… so anything the data says should not violate the ontology
So they could underlie the models of data, messages, documents and so on? Yes: ideally those MODELs need to be informed by
ontologies
© Thomas Beale 2009
What kind of solution do we need?
But wait, there are other dimensions to this problem….we also need:
rules for ACCESS CONTROL IDENTIFICATION of artefacts SIGNING and AUDIT TRAILing of artefacts GOVERNANCE rules, defining development
and dissemination processes, so that QUALITY can be assured
© Thomas Beale 2009
So all this means….?
Firstly, it’s all connected…just like the systems in our requirements…which means the specifications need to be DESIGNED TOGETHER in a single framework
Secondly, there is a difference between ‘standards’ for PARTICULAR MEDIA (e.g. a relational DB, an HL7v2 XML message) and technologies (.Net Winforms, Java Swing), and underlying SEMANTIC MODELS
© Thomas Beale 2009
What we have today
Documentschemas
Messageschemas
GUIdefinitions
Serviceinterfaceschemas
+ terminology
…technically and organisationally disconnected…
Terminologies
SDO SDO SDO ???
???
Standards for particular media
© Thomas Beale 2009
IHE
HL7
IHTSDO
ISO
WHO
CEN
ASTM
PDQ
PMACE
N13606-4
RB
AC
Security
SNOMED CTICDx
Terminology
Services
HSSP
PIX
HISA
RID
XD
S
EN13606-5
Content models EN
13606-3
EN
13606-2
Tem
plat
es
Documents
CDA r2
EN13606-1
CCR
CDA r1
.... and there are a lot of them...
v2 m
essa
ges
v3 messages
Data
types
CCOW
Other
© Thomas Beale 2009
What we need…
Documentschemas
Messageschemas
GUIdefinitions
Serviceinterfaceschemas
TerminologiesLanguage of modelsof content & process
Ontologies
Models of content
Models of interfaces
Models of workflow
A coherentsemantic foundation
Standards for particular media
Generationof concreteimplementationspecifications
Models ofinformation
© Thomas Beale 2009
+ a maintenance mechanism
Historically, official standards organisations have had limited ability to maintain standards
Yet no technical artefact is perfect in its first version - it must have a maintenance path!
In the software world, maintenance means: Users being able to
report issues & observe progress Obtain new releases or fixes
The developers reacting and applying changes Issuing new releases
© Thomas Beale 2009
+ an architectural notion
The platform paradigm: Separation of back-end ‘services’ and front-end
‘applications’ Allows flexible composition of system Allows incremental deployment Avoids vendor lock-in
Also known as ‘SOA’, services-oriented architecture
(But note that many SOA advocates forget to specify the information…)
© Thomas Beale 2009
The platform architecture
Service interface
User interface
Application
KnowledgeRepository
otherService
DataRepository
Application
documentsmodels, ref data
messages
xxx
© Thomas Beale 2009
What standards specify
Application
KnowledgeRepository
otherService
DataRepository
Application
documentsmodels, ref data
messages
xxx
Documentschemas
Messageschemas
GUIdefinitions
Serviceinterfaceschemas
© Thomas Beale 2009
Today’s situation…
Amazingly, many countries entrust the creation of critical e-health standards to delegations attending committee meetings at organisations where: Development is done by ad hoc discussion and voting Unstated or non-existent technical paradigms No design process The ‘developers’ are people with limited or no technical
competence and are self-chosen There is no software validation basis There is no maintenance path There are no reliable delivery timelines
This is paralysing e-health programmes today…
© Thomas Beale 2009
So how should we make ‘standards’?
Documentschemas
Messageschemas
GUIdefinitions
Serviceinterfaceschemas
Create alibrary of semanticdefinitions and tools
Inside an opendevelopmentand testingorganisation
+ transformation tools
…i.e. a ‘machine’ for generating ‘standards’…
© Thomas Beale 2009
What about official standards?
We have to ask the question: are they really needed in e-health?
Let’s look at some existing standards that work: ‘small’ - ISO 8601:2004 – date/time string
standard; widely implemented and used; ‘medium’ – W3C XML-schema – massive use in
industry ‘large’ – IETF standards for the internet;
foundational for modern society
© Thomas Beale 2009
ISO 8601:2004
This standard is needed because: Addresses a fundamental need – to communicate times &
dates Needed within and between all domains, e.g. billing /
defence / e-gov systems
It is successful because: Spec is short, low complexity implementable Very generic, used across many domains Very stable
But… even this simple standard contains errors through not
having been implemented and tested properly first Being an ISO standard, no agile means of maintenance
© Thomas Beale 2009
W3C XML schema 1.0
This standard is needed because: Addresses need for technology-independent
representation of shareable data Very generic, used across many domains
It is successful because: High perceived business value Massive $$ by companies spent over last 10y to
create tools disseminated via use of a few well-known tools,
rather than re-implementation of paper specification It has a maintenance mechanism and path
© Thomas Beale 2009
IETF internet standards
These standards are needed because: Collectively provide basis for information environment of
modern civilisation
They are successful because: Long gestation by engineers / universities Each spec only published when implemented in 2 places Each spec limited in scope, low coupling with others
About 70 full standard RFCs Every addition was compatible with existing specs –
principle of COHERENCE
The details of its operations have changed considerably as it has grown, but the basic mechanism remains publication of draft specifications, review and independent testing by participants, and republication. Interoperability is the chief test for IETF specifications becoming standards. Most of its specifications are focused on single protocols rather than tightly-interlocked systems. This has allowed its protocols to be used in many different systems, and its standards are routinely re-used by bodies which create full-fledged architectures (e.g. 3GPP IMS). - wikipedia IETF article, 2009
© Thomas Beale 2009
What works?
With large numbers of individual specifications that need to work together, the model that has worked in the past is exemplified by W3C and IETF… Individual membership by experts, not national
‘delegations’ Online community – mailing lists, wikis, online
(free) specifications Work done mainly via open source & other
implementations, i.e. by technical developers, not committees
© Thomas Beale 2009
The e-health domain
What does e-health look like as a standards problem?
Many parts, but all related – an ecosystem Multiple terminologies Information models Workflow / process definitions Clinical / domain models Numerous concrete representations – messages,
documents, application screens
This is a similar profile to IETF or the W3C family of standards
© Thomas Beale 2009
What role in e-health for official SDOs?
Documentschemas
Messageschemas
GUIdefinitions
Serviceinterfaceschemas
XXX.org
ISO
CEN
ASTM
HL7
SDOs could approve particular releasesrequired by industry
And should considertransferring most existingactivities and resources into the .org
© Thomas Beale 2009
Key Principles
Coherence: all models and specifications are developed in a SINGLE FRAMEWORK
Filtering: any de jure standard that is to be used is re-engineered to a form 100% consistent with existing specifications
Development paradigm: engineering analysis, design, implementation and validation process is used
Maintenance: the mechanisms and organisational commitment for ongoing maintenance of the framework
© Thomas Beale 2009
Key Principles
Openness: the organisation’s requirements and outputs are always open to inspection
Computable dissemination: use in industry should be via small number of solid implementations and/or formal expressions, not continual re-implementation of paper specifications
© Thomas Beale 2009
Key lessons for governments
‘Choosing standards’ as a way of governments solving semantic interoperability is a fallacy – official standards in e-health are incoherent
The official standards ‘development process’ cannot generate the required outputs because of the committee approach
© Thomas Beale 2009
Key lessons
Instead, an open source .org approach is needed to create meaningful ‘standards’
Official standards bodies should be restricted to approving particular releases of particular specifications as having been quality-assured for use for a defined scope or purpose
© Thomas Beale 2009
Standards Reform in e-health
…means building the .org(s): Only standing committees should exist, for defining
requirements, and for review purposes All other work should be done by technical project groups All IP should be developed in a single coherent technical
framework, consisting of: A published corpus of formal specifications Software implementations Other formal artefacts, e.g. schemas, models Open source tools should be available for parsing, viewing
and editing formal artefacts If it is not formal, it does not compute!
© Thomas Beale 2009
Standards Reform in e-health Development environment including:
Version and release management tools Online issue tracking Wiki
Community environment including: Website, mailing lists, wiki
Development done by recognised experts from industry & academia, organised into structured projects on the basis of capability
Funded by beneficiaries, i.e. governments and industry