Rapid Assessment of Honey Physicochemical Properties by Mid-Infrared Spectroscopy and Chemometric Modeling

This study aims to develop multivariate calibration models for the physicochemical parameters of Brazilian honey samples using mid-infrared (MIR) spectroscopy. A total of 97 samples of honey from Apis melliferabees from different botanical and geographical origins in Brazil were subjected to physicochemical analyses of pH, moisture, total soluble solids, free acidity, lactonic acidity, total acidity, proline, diastatic activity, hydroxymethylfurfural, and MIR. Principal component analysis (PCA) was performed as an exploratory technique to assess the dispersion of the samples across physicochemical parameters. To predict physicochemical parameters of honey from MIR spectral data, spectral interval selection via partial least-squares regression (iPLS) was used. The PCA was performed based on physicochemical parameters, with the samples dispersed according to the quality standards set by Brazilian legislation. MIR spectroscopy, combined with iPLS analysis, allowed the construction of models with high predictive capacity for all parameters evaluated in honey samples. The models presented statistical validation parameters, including correlation coefficients ranging from 0.81 to 0.97, performance ratios for deviation from 1.7 to 4.4, and interval error ratios between 5.7 and 16.3. The results demonstrate the potential of using MIR spectroscopy in conjunction with iPLS for the simultaneous, fast, and nondestructive quantification of physicochemical parameters of honey.