Chemical hazards in rice processing and AI-assisted prevention and control: a review of formation pathways, health risks, and intelligent mitigation strategies

Chemical contamination in rice processing arises from process-induced compounds (acrylamide, furan, heterocyclic amines), storage-related mycotoxins (aflatoxins), and exogenously introduced substances (additives, packaging migrants, polycyclic aromatic hydrocarbons). Long-term excessive intake is associated with hepatotoxicity, neurotoxicity, reproductive disorders, and carcinogenic risk. Conventional control strategies are constrained by cumbersome detection, single-factor analysis, and static risk assessment. AI-driven machine learning offers new opportunities for hazard prediction, rapid detection, process optimization, and dynamic risk assessment by integrating multi-source data and constructing high-dimensional nonlinear models. Key mitigation approaches include optimizing thermal parameters, adopting non-thermal processing technologies, regulating additive usage, and substituting synthetic additives with natural alternatives. However, most AI applications in this domain remain at the laboratory or proof-of-concept stage, facing challenges in industrial scalability, model interpretability, and data standardization. Moreover, the vision of full-chain intelligent regulation-from raw material sourcing to consumption-remains a future aspiration rather than an accomplished reality, as existing implementations are predominantly single-point solutions. This review systematically synthesizes the types, formation pathways, and health impacts of chemical hazards in rice processing, critically evaluates the current capabilities and limitations of machine learning-based approaches, and identifies future research directions. The findings provide theoretical and practical support for the intelligent upgrading of rice processing technology and facilitate the shift from passive hazard response to active prediction and intelligent regulation.