Development of a novel NIR spectroscopy-based chemometric model for sourdough bread authentication

The global popularity of sourdough has increased the need for authentication methods, especially in the absence of regulatory frameworks. Traditional biochemical analyses are often labor-intensive whereas near-infrared (NIR) spectroscopy offers a rapid non-destructive alternative. Here, a NIR-based method was developed to distinguish sourdough-containing breads from those leavened with baker's yeast. Several breads were produced modifying the leavening agent (fresh or dry baker's yeast, biga, type II and type III sourdoughs), the percentage of inclusion, and fermentation time. A three-step data analysis, encompassing spectral preprocessing and a range of multivariate analysis techniques, was performed. To achieve sample classification, a Partial Least Squares Discriminant Analysis model was constructed using eight latent variables. The model yielded a sensitivity of 100 % and a specificity of 89 %. The most significant spectral regions driving the separation corresponded to characteristic functional groups associated with sourdough metabolites. Permutation testing confirmed the robustness and reliability of the model. This study demonstrates, for the first time, the suitability of NIR spectroscopy in sourdough bread authentication. The proposed method enables rapid, non-invasive classification with high accuracy, addressing a critical need for transparency in food traceability and labeling.