Artificial intelligence enabled food quality assessment through digital sensing and explainable analytics

Ensuring food quality and safety has become increasingly challenging due to globalization of food supply chains, rising food adulteration, increasing consumer expectations, and stringent regulatory requirements. Conventional food quality assessment methods are often labor-intensive, destructive, time-consuming, and unsuitable for real-time industrial monitoring. Recent advances in digital sensing technologies, including hyperspectral imaging, biosensors, electronic noses, and Internet of Things (IoT)-enabled platforms, combined with artificial intelligence (AI), machine learning (ML), and deep learning (DL), have emerged as promising solutions for rapid, non-destructive, and scalable food quality assessment. Although numerous studies have reported AI applications in food monitoring, most existing reviews discuss sensing technologies or AI algorithms separately and inadequately address challenges related to scalability, interpretability, sensor heterogeneity, multimodal data integration, industrial deployment, and real-world generalization. Furthermore, limited attention has been given to explainable AI (XAI), edge-AI implementation, ethical considerations, and standardized analytical frameworks. This review provides a critical and integrated overview of AI-driven food quality and safety assessment by examining digital sensing technologies, multimodal data preprocessing, feature engineering, ML/DL architectures, XAI frameworks, and real-time deployment strategies. Current challenges, including overfitting, limited dataset diversity, sensor drift, computational complexity, and regulatory reliability, are critically evaluated. The review further highlights emerging directions, including multimodal sensor fusion, federated learning, edge AI, digital twins, and predictive analytics, which are expected to enable scalable, interpretable, and sustainable food quality monitoring across the farm-to-fork continuum.