Expert system

From: CARLOS QUINONES (carlosq@epm.net.co)
Date: Wed Jan 27 1999 - 14:32:19 EST


Dear Sirs

There is some confusion in the mail list about this topic, I will explain
it in a better way.  I try to find information similar to Fidelio and
Petrushka software from Coulter, but shareware.  Of course, the system can
be complemented with the cluster , discriminant or neural networks
analysis, but this topic is secondary and besides there are excellent
shareware software as Win Mdi (a good idea it would be join the expert
system to Win Mdi).
Other obvious need are the databases that were used for obtaining some
results or some discriminant functions.  Some years ago it was impossible
to supply these databases in a printed journal, but today it is possible. 
Some virtual communities have understood this necessity and, for example,
in StatLib, a web position about statistics algorithms, you can find the
databases that were used in journals and books.  In this way the results
are REPRODUCIBLE.  When you check the bibliography about discriminant
functions, by example, you can find discussions and discussions.  Because
the databases are not available, one research obtains different results
from other, although they use the same statistical procedures.  Then, the
questions begin: Others researches used patients very young? very old? did
they round the numbers?.  Besides, there is too much duplicate work.  If
you have a good idea about a statistical procedure you have to obtain  600
records again.
If someone has these databases it is possible to create a free web position
with these results and many researches can advance faster and cheaper.
The source code of the shareware software can be included and so you avoid
the black box syndrome of this software.
Three important papers about this topic (but until I understand, they are
not a shareware software) are:
-Evaluation of a knowledge based system for interpreting flow cytometric
inmunophenotyping data.  Proceedings of MIB 93. Diamond, L et al.1993. pp
124-128
-Computer assisted bone marrow interpretation:A pattern approach. 
Proceedings of MIB 93. Nguyen, D et al. 1993. pp 119-123 
- Localization of monoclonal B-cell populations through the use of
Kolgomorov-Smirnov D-value and reduced Chi-square contours.  Cytometry. 
Bagwell, C;Loven, EJ;Ault, KA. 1988. pp 469-476

Final note: In hematology there are many programs that were made decade
after decade.  Examples: ANEMIA, Context-I, HEME etc...  If these systems
are 
not commercial or they were abandonated, it would be a good idea that
appeared in a web position.

Best regards

Carlos Quinones



This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:53:01 EST