Abstract
Non-invasive blood glucose monitoring remains a major challenge in biomedical sensing due to strong light scattering in biological tissues, physiological variability, and limited signal stability of existing optical methods.Near-infrared (NIR) spectroscopy has attracted significant interest as a promising approach for continuous and painless glucose monitoring; however, many reported systems remain confined to laboratory conditions and lack sufficient experimental validation.In this study, a compact multispectral non-invasive sensor system based on NIR spectroscopy is developed and experimentally validated.A mathematical model of optical absorption in biological tissues, based on the Beer-Lambert law and implemented in the MATLAB/Simulink environment, was used to identify wavelength regions exhibiting favorable sensitivity-stability trade-offs.Based on simulation results, four operating wavelengths (940, 1050, 1200, and 1350 nm) were selected for sensor implementation.The proposed system integrates near-infrared light-emitting diodes, a photodiode with low-noise amplification, an analog-to-digital conversion stage, and a microcontroller-based data acquisition unit.Experimental validation was performed under both in vitro measurements using aqueous glucose solutions and in vivo measurements conducted on the human earlobe in a transmission configuration.The results demonstrate a strong correlation between optical signal attenuation and glucose concentration (r > 0.95), with a relative measurement deviation not exceeding 5% under controlled experimental conditions.The highest sensitivity was observed at 940 nm, while longer wavelengths (1200-1350 nm) provided enhanced signal stability.Digital signal processing enabled noise reduction of approximately 25-30%, improving measurement reproducibility.Overall, the results confirm the feasibility of the proposed multispectral NIR-based sensor as a proof-of-concept platform for non-invasive glucose monitoring and provide a basis for further optimization and extended experimental and preclinical validation studies.