Re: DM: Data mining in Medicine

[Miguel Artur Feldens]

Hi!

I work in a research project which aims to develop different 
"intelligent
information system" technologies applied to hospital/medical areas. 
My area
is data mining in health-care and hospital administration, currently
focusing on tools and methodology.

In my master thesis, I have developed and applied a data mining tool 
to
discover rules of the forms:

        IF diagnosis X and procedure Y and...
        THEN internment time is 0

Hundreds of rules pointing at the number of days the patients stay in
hospital have been discovered from a General Hospital database, which 
are
useful for tracking certain kinds of anomalies/frauds, for examples.

Rules pointing at the charge reason (patient transference, death..) 
have
been also searched and found. These can be useful for discovering 
clinical
situations with high rates of transference and/or death, which may 
reveal
the need for improvement of certain kinds of resources in each 
hospital
(equipment,personal...).

These are not "medical" applications, but hospital administration and
hospital information system administration applications, but I hope 
these
are interesting for you. These applications where described in my 
master's
thesis (sorry, Portuguese only!), but also appeared on papers (some in
English). Check them in my page: http://www.inf.ufrgs.br/~feldens

This work resulted in the implementation of a tool called AIRA Data 
Mining
Tool, which has been developed in association with a company called 
Hycones
(http://www.hycones.com.br). I guess there's a demo available in this
address...

We are starting a new series of applications in another hospital, 
using
hospital account information to look for associations between patient
information, medical material, procedures, medical doctors, etc. 
Another
potential application in near future is relating clinical situation 
with
surgery time, in order to implement som kind of surgery room 
allocation
protocol, for the non-emergency cases. This is also to improve 
hospital
resource (rooms) allocation.

Best wishes!

Miguel



========================================================================
Miguel Artur Feldens - Universidade Federal do Rio Grande do Sul
feldens@inf.ufrgs.br - http://www.inf.ufrgs.br/~feldens
------------------------------------------------------------------------
Intelligent Databases Group - http://www.inf.ufrgs.br/gpesquisa/bdi
SIDI Project - http://www.inf.ufrgs.br/gpesquisa/bdi/SIDI
UFRGS Knowledge Discovery Group - http://www.inf.ufrgs.br/gikdd
========================================================================