Premium
Expert systems and relevant information
Author(s) -
Zvárová Jana
Publication year - 1999
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/(sici)1099-095x(199907/08)10:4<493::aid-env370>3.0.co;2-#
Subject(s) - computer science , data science , econometrics , mathematics
To support medical decision‐making tasks successfully, we need to know whether the measured features on patients are relevant. In probabilistic expert system, optimal decision‐making is mostly based on the minimization of the average risk (Bayes risk). Therefore, before we try to find the best probabilistic expert system, we should evaluate the medical data using data reduction and constitution procedures. In this paper we describe an information theoretic approach for data reduction and constitution that is also based on the Bayesian criterion of the optimality. The program CORE offers the possibility to estimate information measures of stochastic dependence from medical data and opens the perspective to perform data reduction and constitution in practice. Copyright © 1999 John Wiley & Sons, Ltd.