Sorry to re-launch the debate after more than a while. I have been seized in
the meantime by urgent duties.
I am in general on line with Serge's opinion.
I believe that a numerical model of living system should better be objective
oriented. By objective I mean that the model is being built to help in
solving a particular problem, whether it is a clinical issue or a research
issue.
A "global virtual human" is likely just not to work for several
complementary reasons. To bring up a few:
-the amount of knowledge to integrate is tremendous and this knowledge is
growing. This has two or three consequences which make the task utopian.
E.g., the choice of the level of integration: a phenomenological approach
might be appropriate today and a source of big concern to morrow because new
data have made it wrong to represent "reality"
-the researchers on charge of building the "global virtual human" will
participate most because the project meet their own research objectives than
because they accept to embark in a end-less task; as a consequence, the
outcome of their contribution will probably be biased by their own research
goal (by biased I mean that their choice of details or integration level
will depend on what they expect from their research). Another way of saying
that is: the 17th century Encyclopedists mood is over.
I prefer to see the STEP initiative as dedicated to target a models library
and a modelling tools library.
However, I fully understand that for political reasons we pursue to speak of
"the European physiome project".
Jean-Pierre Boissel

-----Message d'origine-----
De : sintjans
Envoyé : lundi 29 mai 2006 15:47
À : The STEP Action public forum
Objet :re:The Elusive Physiome

Dear all,

>Philip comments seemed provocative, because he has a different
>perspective, more relate to the final clinical application then most
>of us. But of course this perspective is very important for our
>exercise.

I do not see Philip's comments as provocative; I find them rather realistic
(even if I'm not fully agreeing with all of them).

I usually see myselve as somebody between real clinicians and developers. I
spent (with "t" and spend (with "d" enough time with both categories of
professionals to know their respective qualities, shortcomings and
expectations. So, here is my two cents on that particular topic.

>After the conference Philip sent me a small document who summarised
>the positions he expressed during the conference.
>http://www.biomedtown.org/biomed_town/STEP/Reception/step_presentations/The
_Elusive_Physiome.doc?action=download

I read Philip's document (thank you, Philip, to relaunch the debate!).

Like him, I do not believe that the all-inclusive, all-integrative Physiome
(i.e., including all organs/cells/genes at all scales/levels...) would be
truly useful in Clinics today. I also do not believe that Clinicians are
waiting for it. In short, I do not believe that a truly Clinical Physiome or
VPH that could help to solve/process "any" pathological phenomenon is
convincing: you would inject any patient data into the Physiome black box
that would return you some diagnosis report ... the level of complexity
involved is way above our head to achieve such goals in a humanly acceptable
timeframe.

However, I truly believe that a Technological Physiome would be extremely
useful. Such environment would define and offer to developers all (or at
least a large part of it) ressources necessary to build integration tools
around particular clinical problems. These tools would be as various as
data, code, litterature, standards, test cases, benchmarks, legal frames,
etc ... Because this Technological environment would be somehow centralized,
the integration of a particular level/scale with another level/scale would
be straigthforward. Integrating an anatomical system with another would also
become easier if required. In short, with such environment available
integrated modelling would be much easier when requested from the field to
develop a Clinical Physiome-like tool. The larger the available environment
will become the higher the complexity of the system could be. This is very
similar to what Randy Thomas described during the meeting in Brussels.

I also accept Philip's comment whe he said that "extreme complexity over a
wide range of temporal and spatial scales tend to elude appropriate
computational modeling" ....

.. but I'm not sure this is why we should therefore not to attempt to solve
such complexity by computational means. Classical today medicine also shows
high failure rates when dealing with such complexity. Even more when dealing
with predictive analysis of such complexity, certainly in our field (you
probably remember the presentation given at the STEP conference by my boss,
Prof Rooze). Why not giving a chance to computational modeling JUST BECAUSE
of that complexity and the lack of satisfactory results from the fragmented
approach of today Medicine? It might fail, but we'll never know unless we
give technology a chance.

Of course, such system must be built not just on technology but also on
human knowledge ... and this probably one of most delicate aspects of the
Physiome development: how to transport all the knowledge and experience
gathered by Clinicians into usable and useful computer models?

And as Philip rightly said, once available, we should "see models for what
they are, and not passed off as something more ambitious. In the foreseeable
future, at least, each model or group of models will be limited and
approximate in relation to living reality. The strengths of a particular
model or group of models will be applied most effectively if its limitations
are also recognised." I think we all agree with that: we are aiming to
develop a tool, which accuracy must be validated; not a system that would
produce the ultimate truth. The final decision should also be let to the
final user, i.e. the human clinician.

Further comment welcome !

Kind regards,

Serge

*****************************************
Serge VAN SINT JAN, Ph.D.
Associate Professor, Marie Curie Fellow
Department of Anatomy (CP 619)
Université Libre de Bruxelles (ULB)
Lennik Street 808
1070 Brussels - Belgium
Phone: +32-2-555-6325 (-6376)
Fax: +32-2-555-6378
Email: sintjans@ulb.ac.be
LAB website: http://homepages.ulb.ac.be/~anatemb



original:
http://www.biomedtown.org/biomed_town/VPH/StepPublic/step-public/08192379230
25
_____________________________
"The STEP Action public forum"
http://www.biomedtown.org/biomed_town/VPH/StepPublic/step-public
"Biomed Town" http://www.biomedtown.org

Although I agree with most have been written, I would like to stress
a few points, so as to remove an potential source of confusion.

After the consensus we had in Brussels the idea of building global
virtual human totally disappeared from our horizon (assuming it even
existed). So let me say it again: the VPH is NOT a global model.

This said the VPH cannot be limited to a 'collection' of models. As
agreed in Brussels the VPH will be:

- Descriptive (which means it is formed by collections of data such
as medical imaging, biomedical signals, etc.)

- Predictive (which means it is formed by collections of models)

- Integrative (which means it has to expose an organisation that
makes possible at any time to combine data and models from disparate
sources, covering disparate scales, and generated with disparate
methods)

Regards

Marco Viceconti



>Sorry to re-launch the debate after more than a while. I have been seized in
>the meantime by urgent duties.
>I am in general on line with Serge's opinion.
>I believe that a numerical model of living system should better be objective
>oriented. By objective I mean that the model is being built to help in
>solving a particular problem, whether it is a clinical issue or a research
>issue.
>A "global virtual human" is likely just not to work for several
>complementary reasons. To bring up a few:
>-the amount of knowledge to integrate is tremendous and this knowledge is
>growing. This has two or three consequences which make the task utopian.
>E.g., the choice of the level of integration: a phenomenological approach
>might be appropriate today and a source of big concern to morrow because new
>data have made it wrong to represent "reality"
>-the researchers on charge of building the "global virtual human" will
>participate most because the project meet their own research objectives than
>because they accept to embark in a end-less task; as a consequence, the
>outcome of their contribution will probably be biased by their own research
>goal (by biased I mean that their choice of details or integration level
>will depend on what they expect from their research). Another way of saying
>that is: the 17th century Encyclopedists mood is over.
>I prefer to see the STEP initiative as dedicated to target a models library
>and a modelling tools library.
>However, I fully understand that for political reasons we pursue to speak of
>"the European physiome project".
>Jean-Pierre Boissel


--
--------------------------------------------------
MARCO VICECONTI, PhD (viceconti@tecno.ior.it)
Laboratorio di Tecnologia Medica tel. 39-051-6366865
Istituti Ortopedici Rizzoli fax. 39-051-6366863
via di barbiano 1/10, 40136 - Bologna, Italy

Tiger! Tiger! Burning bright in the forest of the night,
what immortal hand or eye could frame thy fearful symmetry?
--------------------------------------------------
Opinions expressed here do not necessarily reflect those of my employer