An Efficient Attribute Ordering Optimization in Bayesian Networks for Prognostic Modeling of the Metabolic Syndrome

The metabolic syndrome has become a significant problem in Asian countries recently due to the change in dietary habit and life style. Bayesian networks provide a robust formalism for probabilistic modeling, so they have been used as a method for prognostic model in medical domain. Since K2 algorithm is influenced by an input order of the attributes, optimization of BN attribute ordering is studied. This paper proposes an evolutionary optimization of attribute ordering in BN to solve this problem using a genetic algorithm with medical knowledge. Experiments have been conducted with the dataset obtained in Yonchon County of Korea, and the proposed model provides better performance than the comparable models.