An adaptive binary PSO to learn bayesian classifier for prognostic modeling of metabolic syndrome

Satchidananda Dehuri, Rahul Roy, Sung-Bae Cho

Abstract: The metabolic syndrome is a combination of medical disorders that have become a significant problem in Asian countries due to the change in lifestyle and food habits. Thus a prognostic model can help the medical experts in diagnosis of the disease. Learnable Bayesian classifier by Adaptive Binary Particle Swarm Optimization (ABPSO) provides a robust formalism for probabilistic modeling that can be used as a predictive tool in medical domain. In this paper, we adopt an ABPSO for adapting the weights of the learnable Bayesian classifier that provides a maximum prediction accuracy and can exhibit an improved capability of removing spurious or little important attributes and help the medical experts in identifying the basis for the disease. Experiments have been conducted with the dataset obtained in Yonchon Country of Korea, and the proposed model provides better performance than the other models.