Principal Component Analysis and Near-Infrared Spectroscopy as Noninvasive Blood Glucose Assay Methods.

Abstract

In this paper, a new model is presented for estimation of the blood glucose level from the measured near-infrared absorbance. The model has been developed in such a way that the regression coefficients of this linear relation have been approximated by considering only the molar absorptivity of the glucose and the obtained coefficients have been utilized to estimate the blood glucose levels from the measured absorbances. The estimation of the blood glucose concentrations by this blind approach exhibited an acceptable accuracy in comparison to the more accurate principal components regression method. The blood sample absorbances have been measured using a Fourier transform infrared device while the blood glucose levels have been determined by a commercial finger-prick glucometer device.