In this chapter, we will present data mining techniques for discovering knowledge from Scoliosis database in the medical domain. Two kinds of knowledge, namely causal structures and rule knowledge, are learned. We employ Bayesian networks to represent causal structures. These networks are capable of depicting the causality relationships among the attributes. Rule knowledge captures interesting patterns and regularities in the database. We develop discovery methods based on Evolutionary Algarithms. Evolutionary Algorithms simulate the natural evolution to perform function optimization and machine learning. In particular, our approach for discovering causality relationships is based on Evolutionary Programming which learns Bayesian network structures. For rule learning, we apply Generic Genetic Programming to discover the rule knowledge. We have discovered new knowledge about the classification of Scoliosis and about the treatment. We demonstrate that the data mining process helps clinicians make decisions and enhance professional training.
|Title of host publication||Medical data mining and knowledge discovery|
|Editors||Krzysztof J. CIOS|
|Place of Publication||United States|
|Publisher||Springer-Verlag GmbH and Co. KG|
|Number of pages||37|
|Publication status||Published - 2001|
WONG, M. L., LAM, W., & LEUNG, K. S. (2001). A medical data mining application based on evolutionary computation. In K. J. CIOS (Ed.), Medical data mining and knowledge discovery (pp. 281-317). Springer-Verlag GmbH and Co. KG.