A New Kernel Extreme Learning Machine Framework for Somatization Disorder Diagnosis

In this paper, we present a novel predictive model based on the kernel extreme learning machine (KELM) to predict the somatization disorder. Since the classification performance of KELM is largely affected by its two parameters, it is necessary to set two optimal parameters for it to ensure high prediction accuracy. In order to improve the accuracy of the prediction model, a new optimization strategy is used to optimize the parameters of KELM. The new optimization strategy adopted grey wolf optimization algorithm to generate high-quality initial populations for moth-flame optimization algorithm, called GWOMFO. The effectiveness ofGWOMFO was first verified on the ten classic benchmark functions. The results show that the GWOMFO has provided consistently better results than other competitive algorithms. This reveals that high-quality initial populations can significantly improve the global search ability and convergence speed of search agents. Furthermore, the proposed GWOMFO-based KELM model was compared with other models, including a model based on GWO (GWO-KELM), a model based on MFO (MFO-KELM), a model based on genetic algorithm (GA-KELM), a model based on grid search method (Grid-KELM), a random forest, and the support vector machines, on the somatization disorder dataset. The simulation results show that the developed framework cannot only achieve higher prediction accuracy than other models but also has better robustness.