Machine Learning-Driven Optimization for Digital Transformation in Non-thermal Food Processing

Abstract Non-thermal food processing has opened up new space and has emerged as a promising alternative to conventional thermal methods of food processing. These foods meet the growing consumer demands for high-quality, convenient, and minimally processed foods. The idea of proposing a machine learning (ML) strategy for finding the optimum process parameters and kinetics in food processing applications is new and challenging, but this new innovative approach requires considerable scientific effort. This review presents the applications of ML in the optimization of non-thermal food processing technologies such as high-pressure processing (HPP), pulsed light (PL), ultrasound (US), pulsed electric fields (PEF), cold plasma (CP), and irradiation (IR). These technologies have exhibited conspicuous advantages with respect to microbial inactivation, preservation of food quality, and environmental sustainability. Integration of ML with non-thermal technologies will enable better control and monitor in real time and optimize critical parameters such as pressure, frequency, and treatment duration. While numerical models have conventionally been used successfully for process optimization, ML provides better adaptability by identification of complex nonlinear relationships in food systems for more accurate prediction and adjustment. The key takeaways of this paper lie in the ML-driven monitoring system, integrated sensors, and real-time data accumulation in response to enhancing process efficiency with dependency natures inherently presented by food matrices. Further development of ML models, apparatus collection, and intelligent systems is expected to yield non-thermal food processing methods with enhanced sustainability, safety, and quality.