On Sat, 16 Dec 2000 23:15:25 Joseph Dal Molin wrote:
>Posted on behalf of Adrian Midgley
>> > Although I can appreciate the benefits of having an intermediate
>> common representation, I just don't see how it is absolutely
>> necessary.
>>
>> Absolute, in theory, no... but in the real world, yes, essential.
>>
>> Wayne did the proper explanation, can I just throw in the minimalist
>> one:-
>>
>> n-1 vs n**2
>>
>> <grin>
>> --
>> Midgley
Hi Adrian,
As a minimalist myself, I am most impressed by your very accurate distillation of
these complex issues to the n-1 vs. n**2 difference. You are most likely correct that
this is at the root of all efforts to establish a common intermediate schema that will
be the mother of all common intermediate schema.
You are also correct that n-1 is less than n**2. However, my argument is precisely
that there are other considerations other than n-1 vs. n**2. Is it possible that the
prevailing choice of the "n-1" solution (even at the expense of other trade-offs) may
be responsible for some of the hard to solve problems in health data interchange? I
think it is helpful to re-examine current solutions and challenge their assumptions.
Let me list some observations for your consideration:
1) The assumption of a fully connected graph is not supported. Most interchanges are
local in nature. A given patient is unlikely to be seen at all clinics in the world -
therefore, a solution that assumes such a scenario is unlikely to be optimal. Thus, I
doubt that we will ever approach n**2. The implication is that when calculating the
cost/benefit of intermediate vs. non-intermediate schema architecture, it is
misleading to assume that all systems must interchange all schema with all other
systems.
2) The cost of instituting a "superset" intermediate schema is unknown (but at least
more than 10*my annual salary <grin>). As I previously raised in reply to Wayne, it
is known to be hard and perhaps impossible to produce such a schema.
3) n-1 can be approached without a common schema if some of the mediators can be
reused (A==>C is A==>B + B==>C).
4) A system that is based on plug-and-play mediators has never been implemented, to my
knowledge. Perhaps it could perform adequately and offer advantages not available to
current systems. If not, such a system (e.g. OIO) could be easily constrained to
operate through a standard/reference schema - thus becoming a traditional n-1 system.
How does that sound???
Andrew
---
Andrew P. Ho, M.D.
OIO: Open Infrastructure for Outcomes
www.TxOutcome.Org
Assistant Clinical Professor
Department of Psychiatry, Harbor-UCLA Medical Center
University of California, Los Angeles
Join 18 million Eudora users by signing up for a free Eudora Web-Mail account at
http://www.eudoramail.com
