Noninvasive Monitoring of Blood Glucose With In Vivo Raman Spectroscopy.

Abstract

Non-invasive glucose monitoring using Raman spectroscopy with 830 nm excitation presents a promising alternative to traditional fingerstick methods for diabetes management research. An integrated in vivo Raman system enables transcutaneous glucose detection and has demonstrated robust performance in oral glucose tolerance tests (OGTT), validating its reliability. Inter-subject correlation between spectral features and glucose concentration was addressed by the intensity of the fingerprint peak (I₁₁₂₅), peak intensity ratio (I₁₁₂₅/I₁₄₄₅), and the spectral area ratio (S₁₁₂₅/S₁₄₄₅), whose correlation coefficient (R) was 0.9266, 0.8946, and 0.9061, respectively. A partial least squares regression (PLSR) model was also adopted for quantitatively bridging the measured Raman spectral information and the actual glucose concentration, showing reliable predictive performance within a wide glucose concentration range of 82.8 to 180 mg/dL. This work demonstrates promising feasibility for in vivo transcutaneous Raman-based glucose monitoring, laying a foundation for subsequent technique transformation in the field of diabetes management and personalized health monitoring.