Selection of singular spectrum analysis components via empirical mode decomposition for extracting information for noninvasive blood glucose estimation system

This paper proposes a method for selecting the singular spectrum analysis components via the empirical mode decomposition approach for extracting the useful information for noninvasive blood glucose estimation systems. To perform the grouping, the total number of the groups of the singular spectrum analysis components is equal to the total number of the intrinsic mode functions. First each normalized singular spectrum analysis component is compared to each normalized intrinsic mode function. Second, the singular spectrum analysis component is assigned to the group corresponding to the intrinsic mode function having the highest correlation coefficient. Third, all the singular spectrum components belong to the same group are summed up together. This technique is applied to extracting the useful information for noninvasive blood glucose estimation systems. In particular, the measured signal is decomposed into a number of components via both the singular spectrum analysis approach and the empirical mode decomposition approach. After applying our proposed grouping method to obtain the singular spectrum analysis components, the obtained components enjoy the advantages of both the singular spectrum analysis approach and the empirical mode decomposition approach. Computer numerical simulations are performed on the practical measurements. The results show that more robust information can be found in the obtained singular spectrum analysis components.