Factors affecting the accuracy of non-invasive blood glucose measurement by short-wavelength near infrared spectroscopy in the determination of the glycaemic index of foods

Non-invasive techniques for the analysis of blood glucose are desirable not only for clinical use but also for the determination of the glycaemic index (GI) of foods. Factors that affect the accuracy of non-invasive blood glucose measurements by near infrared (NIR) spectroscopy were identified and examined to establish the optimal conditions for the analysis. Calibration models were obtained by means of partial least squares regression. Several experiments revealed that a good calibration model could be obtained using spectra (7001100 nm) acquired from the palm of the hand with a large interactance probe fixed to a table provided that spectral acquisition and blood glucose analysis are performed simultaneously. The accuracy of the calibration model developed is clinically acceptable; the coefficient of multiple determination was 0.85 and the standard error of cross-validation was 9.7 mg dL(-1). An analysis of a model solution confirmed that NIR spectroscopy in the short-wavelength region 700-1100 nm is capable of measuring glucose concentrations of the order of 0.01% (10 mg dL(-1)) in blood. The determination of the GI of foods by using the present method will be examined in the near future.