Vibrational spectroscopy and machine learning in bakery production: Toward digital twin integration

Bakery processing is a complex, multi-stage process that involves selecting raw materials, mixing, proofing, and baking. These stages encompass numerous interactions between formulation components and fermentative microorganisms, which critically influence the chemical and physical properties of the dough matrix and the quality of the final product. Advancing the understanding of these dynamic processes requires systematic, real-time monitoring. Conventional quality assurance methods are often destructive, labor-intensive, and time-consuming. Vibrational spectroscopic techniques are emerging as rapid, cost-effective, and non-destructive alternatives for process and quality control across bakery production stages. This review synthesizes recent advances in the application of vibrational spectroscopic techniques, combined with machine learning, for quality assessment of various parameters throughout the breadmaking process. The discussion covers spectroscopy principles, modeling, with particular emphasis on near-infrared (NIR) spectroscopy and hyperspectral imaging (HSI). The concept of a digital twin is introduced as a future direction for intelligent bakery systems. A digital twin driven by vibrational spectroscopy and machine learning could enable real-time simulation, adaptive control, and cross-stage coordination. Future research should focus on developing integrated sensing and modeling platforms that support predictive and resource-efficient bakery processing aligned with Industry 4.0 principles.