health-ont
[Top] [All Lists]

Re: [health-ont] Progress towards our Ontolog RFI

To: "Health-Ont" <health-ont@xxxxxxxxxxxxxxxx>
From: Adam Pease <adampease@xxxxxxxxxxxxx>
Date: Sun, 16 Jan 2005 19:36:34 -0800
Message-id: <6.2.0.14.2.20050116190238.02c3cf98@xxxxxxxxxxxxxxxxxx>
Folks,
   Here's my take on the first three questions in the RFI.    (01)

Adam    (02)


1.The primary impetus for considering a NHIN is to achieve interoperability 
of health information technologies used in the mainstream delivery of 
health care in America. Please provide your working definition of a NHIN as 
completely as possible, particularly as it pertains to the information 
contained in or used by electronic health records. Please include key 
barriers to this interoperability that exist or are envisioned, and key 
enablers that exist or are envisioned. This description will allow 
reviewers of your submission to better interpret your responses to 
subsequent questions in this RFI regarding interoperability.    (03)

   A key aspect of successful interoperability is shared meaning.  The 
current state of practice in information systems is to develop interfaces 
between pairs of systems, where the interface strictly addresses the 
information of concern only to that pair.  Efforts to employ common syntax 
are a necessary start, but do not solve the interoperability problem.
   Much effort is currently being spent on redefining data models in XML 
syntax.  Sharing a standard syntax can greatly improve the ease of 
integration across a large infrastructure.  However, shared syntax is not 
sufficient without shared meaning.  For that reason, an increasing amount 
of effort in the IT industry is being spent on developing shared XML 
schema.  However, that too is insufficient, since such efforts tend to 
address constrained domains that are likely to break when requirements 
expand.  Such efforts, by being specific to particular industries, also 
fail to reuse the best and most general information models that exist 
across domains.  Lastly, even when schemas are shared and terms are 
carefully defined in English (or another human language), the meaning of a 
term is only as clear as its English definition.  English sentences are 
invariably subject to ambiguity, vagueness and issues of how context 
affects meaning and interpretation.
   The solution we envision has several parts, to address each of the key 
barriers just described.  We believe that shared meaning is a critical 
aspect of interoperability.  To have shared meaning we must have not only a 
common syntax, but also a common vocabulary.  That common vocabulary should 
be defined in terms of the broadest and most general foundation concepts, 
in order to have robustness when requirements change or expand.  The common 
vocabulary must be defined in a formal and computable language, so that 
interpretation of meaning can be supported by computers, and not subject to 
human interpretation.  The common vocabulary should reuse as much general 
conceptual infrastructure as possible, in order to amortize costs over the 
widest possible set of interfaces, and to ensure the quality that comes 
from repeated use and testing of common information products.
   Specifically, we believe that a formal upper ontology, defined in logic, 
is a necessary component in achieving large scale interoperability.  We 
envision a hierarchy of ontologies that build from a common semantic 
foundation.    (04)


2.What type of model could be needed to have a NHIN that: allows widely 
available access to information as it is produced and used across the 
health care continuum; enables interoperability and clinical health 
information exchange broadly across most/all HIT solutions; protects 
patients' individually-identifiable health information; and allows vendors 
and other technology partners to be able to use the NHIN in the pursuit of 
their business objectives? Please include considerations such as roles of 
various private- and public- sector entities in your response.    (05)

A fully formal and logical model will help to ensure that clients and 
participants in a common health network will be able to understand how to 
connect to the common information model.  The current practice often 
involves discussion with model "experts" or authors in order to understand 
fully what is meant by each term.  A formal model will help ensure that all 
information and context is fully explicit, and nothing is left unstated and 
only in the head of the designer.
   We believe that a completely open information model is also a 
necessity.  Vendors must be able to use the common model without fee in 
order to spur adoption.  The open source model will mean that academic 
institutions will be able to employ student resources to assist in 
infrastructure development, research and use, which would be much less 
likely if models were costly and proprietary.
   While a formal, common information model does not directly address 
privacy issues, the use of a general upper model will ensure that many 
real-world aspects that have an impact on privacy can be successfully 
modeled, and therefore be taken into account in an integrated system that 
handles privacy concerns in both storage and access.    (06)


3.What aspects of a NHIN could be national in scope (i.e., centralized 
commonality or controlled at the national level), versus those that are 
local or regional in scope (i.e., decentralized commonality or controlled 
at the regional level)? Please describe the roles of entities at those 
levels. (Note: "national" and  "regional" are not meant to imply federal or 
local governments in this context.)    (07)

We believe that a hierarchy of information models will be needed.  The 
upper ontology will be centralized and standard.  We believe that there 
will be significant benefit in having some standardized, centralized body 
of health ontology information as well.  If there is a proliferation of 
vocabularies with overlapping semantics, we will not be able to achieve the 
goal of interoperability.
   There have been several barriers to large common information models in 
the past    (08)

- The language in which models have been stated is insufficiently 
expressive, leading to "overloading" of language aspects for uses they were 
not intended
- The language is insufficiently explicit, leading to a "grab bag" of 
models which are not provably overlapping, but which in fact are mutually 
redundant or inconsistent
- The language is insufficiently formal, requiring human experts to 
understand the implicit definitions of terms in the model in order to 
assess whether model additions are redundant or contradictory
- The model doesn't inherit from a common upper model, leading to 
duplication of general purpose concepts, and a proliferation of 
incompatible upper level concepts which each build in different simplifying 
assumptions.    (09)

At 12:33 PM 1/16/2005, Bob Smith wrote:
>Hello,
>
>We are making significant progress with the Managerial and Organizational 
>questions, and hope to see the questions from the other members of our project.
>
>Please upload or eMail me your responses (best in Word, as that is the 
>format Dr. Brailer requested).
>
>If you have questions, please call me at 714 536 1084
>
>Thanks, and looking forward to your responses.
>
>Bob Smith    (010)

----------------------------
Adam Pease
http://www.ontologyportal.org - Free ontologies and tools    (011)


 _________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/health-ont/
Community Files: http://ontolog.cim3.net/file/work/health-ont/NHIN-RFI/
To  Post: mailto:health-ont@xxxxxxxxxxxxxxxx
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?NhinRfi    (012)
<Prev in Thread] Current Thread [Next in Thread>