Ultraviolet–visible spectral characterization and ANN modeling of aqueous sugar solutions: Clinical and environmental perspectives

The characterization of aqueous sugar solutions using optical techniques offers a non-invasive, rapid, and reagent-free approach for concentration monitoring in both analytical and environmental contexts. In this study, aqueous D-glucose solutions at concentrations of 0.1, 0.2, 10, 20, and 40 g/mL were analyzed using an ultraviolet–visible–near-infrared spectrophotometer across the 200–1020 nm wavelength range. Although glucose exhibits inherently low absorbance in this spectral domain due to the absence of strong chromophoric groups, measurable trends were observed—particularly in the ultraviolet region below 400 nm—consistent with theoretical expectations based on the Beer–Lambert law. Absorbance intensity increased consistently with glucose concentration, and while no sharp absorbance peaks were detected, subtle spectral variations encoded sufficient information to enable computational modeling. A feed forward artificial neural network was trained on the full spectral dataset and demonstrated high predictive accuracy, achieving a correlation coefficient exceeding 0.98. These findings underscore the potential of integrating ultraviolet–visible spectroscopy with machine learning techniques for real-time, label-free detection of glucose and similar analytes. The approach not only supports the development of fast and accurate monitoring systems in clinical and industrial settings but also lays the groundwork for future research involving more complex sugar matrices and environmentally relevant applications.