Improvement of partial least squares models for in vitro and in vivo glucose quantifications by using near-infrared spectroscopy and searching combination moving window partial least squares

The aim of this study is to develop optimum partial least squares (PLS) models for glucose quantifications in bovine serum samples and human skin by using near-infrared (NIR) spectroscopy and a chemometrics algorithm named searching combination moving window partial least squares (SCMWPLS). This chemometric method allows one to search for either an optimized spectral region or an optimized combination of the informative spectral regions selected by an early proposed wavelength interval selection method called moving window partial least squares regression (MWPLSR). By using SCMWPLS, we can find the optimum spectral regions for developing efficient PLS models of glucose in the bovine serum samples and the human skin. In the present study, we evaluate the performance of SCMWPLS with the two different NIR data sets, the in vitro NIR spectra of bovine serum samples and the in vivo NIR spectra of human skin.