Abstract: The proposed Bacterial foraging optimization algorithm (BFO) based on support vector machine (SVM) presents an effective intelligent early-warning model (IBFO-SVM) for the somatization disorder severity prediction for Community Correction patients. In IBFO-SVM, opposition-based learning strategy is introduced into the BFO to increase the diversity of the bacterial population and uniformly distributed as far as possible for initial population, which improves the convergence rate of BFO and reduces the probability which falls into the local optimal solution. The proposed improved bacterial foraging optimization algorithm (IBFO) is used to determine the two optimal parameters of SVM (penalty coefficient and kernel width). For comparison purpose, the original BFO, genetic algorithm (GA), particle swarm optimization (PSO) has also been proposed to optimize the parameters of SVM. In this study, IBFO-SVM, BFO-SVM, GA-SVM and PSO-SVM model were compared by 10-fold cross-validation method in psychological evaluation data. The experimental results demonstrate that the proposed IBFO-SVM method has better performance in predicting the severe somatization disorder and mild somatization disorder in terms of classification accuracy, Mathews correlation coefficient (MCC), sensitivity and specificity than other methods.

Key words: computer application, somatization disorder, support vector machine, opposition-based learning, bacterial foraging algorithm