An Intelligent Diagnosis to Type 2 Diabetes Based on QPSO Algorithm and WLS-SVM

Abstract

The pre-diagnosis to type 2 diabetes, and the effective prophylaxis and treatment of its complication is to be worthy paying attention to. So an intelligent diagnosis based on quantum particle swarm optimization (QPSO) algorithm and weighted least squares support vector machines (WLS-SVM) is presented, which can overcome the disadvantage of large sample data, slow model-building and rather large deviation in real-time diagnosis. The detailed improvement of the method is to build a mixed kernel function instead of the single one, add self adapting weights, and solve the linear system of equations in the training model of the WLS-SVM with QPSO algorithm, which can increase the performance of diagnostic model. Applied the method in type 2 diabetes, it shows that the velocity of the model-building is quick and the diagnosis accuracy is high, and the result of the improved WLS-SVM is superior to the improved BP algorithm, LM algorithm neural network and the single-kernel function SVM.