Publications

Guest editor: Krzysztof J. Cios.

Data mining and knowledge discovery in medical databases are not
substantially different from mining in other types of databases. There are
some characteristic features, however, that are absent in non-medical data.
The most pronounced is the fact that more and more medical procedures
employ imaging as a preferred diagnostic tool. Thus, there is a need to
develop methods for efficient mining in databases of images, which is not
only different, but also more difficult, than mining in numerical
databases. Other features are security and confidentiality concerns. Still
another one is the fact that the physician's interpretation of images,
signals, or any other clinical data, is written in unstructured English
that is very difficult to mine in.

The articles in this special issue address many of the above-mentioned
topics. Some concentrate more on data mining methods, while other papers
describe the entire knowledge discovery process; all use medical data but
of widely differing sizes.   The first article describes key issues
encountered in data mining, outlines six-step knowledge discovery process
(understanding medical problem domain, understanding the data, preparation
of the data, data mining, evaluation of the discovered knowledge, and using
the discovered knowledge), and uses it on a database of bull's eye SPECT
images of the heart.  The second article addresses an important problem of
consistent knowledge discovery in medical diagnosis. The third paper
describes how genetic programming can be used for generating diagnostic
rules for chest pain diagnosis.  The next paper also uses evolutionary
algorithms for discovering knowledge from medical databases. The fifth
article describes a system for discovery of positive and negative knowledge
in clinical databases using the theory of rough sets, one of the newest
data mining methods. Another paper addresses the problem of temporal
pattern discovery in course-of-disease data. The following article
describes the use of inductive logic programming for generating diagnostic
rules from quantitative MRI data to discriminate between meningiomas and
astrocytomas. The last paper uses the six-step knowledge discovery process
on a database of 613 patients, containing cardiac SPECT images and other
clinical information, to generate diagnostic rules classifying perfusion
defects of the left ventricle.

Krzysztof J. Cios
a.k.a. Krys
Professor and Chair
Computer Science and Engineering Department
University of Colorado at Denver
Campus Box 109
Denver, CO 80217-3364

phone: 303-556-4083
fax: 303-556-8369
kcios@carbon.cudenver.edu