Human Blood Glucose Measurement Based on Infra-Red Spectroscopy

A predictive algorithm based on Partial least square (PLS) regression has been developed and optimized for measuring blood glucose levels of in-vitro human blood samples using their infra-red transmission spectrum. The labels - the accurate glucose concentrations - of each blood sample was measured using commercial invasive glucometers. The samples transmission spectra were obtained using a standard Fourier-Transform Infra-Red (FTIR) spectrometer. Using the developed algorithm, the results of the FTIR spectroscopy of the samples were processed and analyzed to find the glucose concentration in the samples. The obtained glucose concentrations by the algorithm were in good agreement with the results obtained by the standard glucometers, with the mean absolute error of 7.5 mg/dL. This error is in the range of the acceptable errors for an available commercial invasive meter.