Blood glucose sensor implant using NIR spectroscopy: Preliminary design study

We previously proposed in [1] a novel standalone fluorescence measurement device that can identify fluorophore substances in a light conducting environment. In this paper, we adapt this work to the particular problem of predicting blood glucose levels based on near-infrared (NIR) absorption spectroscopy. We have focused this investigation on the combination (5000-4000 cm^-1) and first-overtone (6500-5500 cm^-1) spectral bands known to be dominated by glucose absorption information. An array of electrically pumped vertical-cavity surface-emitting lasers (VCSELs) is used to cover the two selected spectral bands; with each laser diode emitting light at a specific wavelength. A Quantum well infrared (QWI) photodetector operating in the 1.2-2.5 μm range is used for optical power detection. Then, an artificial neural network (ANN) based model is used to determine the glucose concentrations from the obtained blood absorption spectra.