Prediction of severe brain damage outcome using two data mining methods

Author

Grzymala-Busse, Jerzy W. ; Hippe, Z.S. ; Mroczek, Teresa ; Bucinski, A. ; Strepikowska, A. ; Tutaj, Andrzej

Author_Institution

Dept. of Comput. Sci., Univ. of Kansas, Lawrence, KS

fYear

2008

fDate

25-27 May 2008

Firstpage

585

Lastpage

590

Abstract

In this paper we report our results on prediction of the Glasgow Outcome Scale for patients affected by severe brain damage. We used two data mining methods: the LEM2 rule induction system and the BeliefSEEKER system generating belief networks. Additionally, the original data set, with missing attribute values and numerical attributes, was mined by the MLEM2 system (a modified version of LEM2). Though our results show that the rule set induced by LEM2 is worse than the rule set obtained by conversion of a belief network generated by the BeliefSEEKER, it is possible to simplify the LEM2 rule set to accomplish similar results.

Keywords

belief networks; data mining; medical computing; patient diagnosis; BeliefSEEKER system; Glasgow outcome scale; LEM2 rule induction system; MLEM2 system; belief networks; data mining methods; severe brain damage outcome prediction; Artificial intelligence; Computer science; Data mining; Expert systems; Induction generators; Information management; Information technology; Machine learning; Nervous system; Set theory; Belief networks; BeliefSEEKER system; Glasgow Outcome Scale; LEM2 rule induction system; severe brain damage;

fLanguage

English

Publisher

ieee

Conference_Titel

Human System Interactions, 2008 Conference on

Conference_Location

Krakow

Print_ISBN

978-1-4244-1542-7

Electronic_ISBN

978-1-4244-1543-4

Type

conf

DOI

10.1109/HSI.2008.4581506

Filename

4581506