by
Christoph Schommer,
IBM German Development Laboratory
The course is dedicated to people who are interested in practical data mining, not only on theoretical aspects. The first part of the tutorial will present a general positioning of data mining in medicine, and address and information retrieval and data mining in a generic context. Then it will discuss in more detail the basics of data mining, including its definition, process, the economic issues related to the application of such methodology and various techniques being used, including an overview of a specific tool being used. The second part of the tutorial will be a practical session, overviewing two medical applications, a live demo of applying data mining on them, followed by an open discussion on the results achieved.
by
Jeremy Wyatt,
Knowledge Management Centre, University College, London,
Doctors rely on at least 2 million facts and clinical knowledge doubles every 20 years. Knowledge management means recognizing the importance of knowledge and mobilizing it in forms that the professional can apply. To understand how this concerns to medical knowledge and AI, this tutorial will review published studies about the clinical use of a range of technologies and techniques, includiing: Internet resources such as content providers, search engines and portals, Decision support, reminder and alerting systems, Practice guidelines and protocol-based care systems, Person-to-person knowledge exchange such as continuing education courses and secondments, Electronic libraries and bibliographic databases. Every participant will receive a free copy of the book on which the tutorial is based, "Clinical knowledge and practice in the information age" (RSM Press, March 2001).
by
Gabriela Guimaraes,
Department of Computer Science Universidade Nova de Lisboa Portugal
An introduction to special neural networks with unsupervised learning will be given, in particular to the Self-Organizing Map (SOM) [Kohonen 82], and its successful applications in medicine. Attendees will obtain a strong backgroung in SOMs and unsupervised learning, in pattern discovery with SOMs, as well as in their applications in medicine. Since temporal applications become more and more important in medicine, a strong emphasis on SOMs for time series analysis will be made. All main mathematical and statistical concepts will be introduced. The main focus, however, will be on the applications.
by
Dan Steinberg,
Salford Systems, San Diego, CA, USA
This relatively non-technical tutorial will develop the key MARS concepts and illustrate how to get the most out of MARS. You will learn: what's under the hood, i.e., how does MARS' engine work?, how to plan and execute a MARS analysis, how to guide runs with key control parameters, what types of data errors can cause MARS difficulties, how to interpret interaction terms, and, how to use MARS to improve an existing model.