Prediction Models for Hyperglycemia Concentrations of Cardiac Conduction System

Sinoatrial node field potential, which is highly sensitive to high glucose, is very suitable to be the parameter for estimating the hyperglycemia concentrations of the environment. In this paper, prediction models are built by using partial least squares (PLS), support vector machine (SVM), back propagation neural network (BPNN) and Takagi-Sugeno (T-S) model respectively to predict the hyperglycemia concentrations of sinoatrial node field potential. Meanwhile, the prediction results of the four models are compared. The results show that the predictive abilities of SVM, BPNN and T-S models are at the same level and the capabilities are the highest when the concentration of glucose is greater than 30mM. PLS is not suitable to predict the high glucose concentrations.