Oct 2026· Food Research International· Vol 242 Pt 5, pp.
120168
· 0 citations· 66 references
Medicine
Abstract
The widespread use of pesticides in modern agriculture has substantially improved food production while raising serious concerns regarding contamination and food safety. Considerable scientific effort has focused on improving methods for monitoring and detecting pesticide residues in food to enhance safety assurance. Although chromatography coupled with mass spectrometry remains the gold standard, its routine application is constrained by high costs, labor-intensive sample preparation, and prolonged analysis times. Recent advances in spectroscopic techniques, including surface-enhanced Raman spectroscopy (SERS), Raman spectroscopy, hyperspectral imaging (HSI), and near-infrared (NIR) spectroscopy, offer promising non-destructive, rapid, and sensitive alternatives for pesticide residue detection across diverse food matrices. When integrated with machine learning (ML), these approaches further improve predictive accuracy and analytical robustness. This review synthesizes recent advances in ML-assisted spectroscopic approaches for pesticide residue detection across diverse food matrices, with emphasis on analytical performance, preprocessing strategies, feature engineering, and model selection. Convolutional neural networks (CNNs), support vector machines (SVMs), random forests (RFs), and ensemble learning methods are increasingly used to improve classification and quantitative prediction. Across the reviewed studies, analytical performance was generally strong, with high classification accuracies, while the lowest reported detection limit was achieved using a SERS-CNN platform. Despite these advances, key limitations remain, including reliance on laboratory-spiked samples, small dataset sizes, matrix interference, inconsistent validation strategies, high computational demands associated with high-dimensional spectral data, and limited field validation. Future directions should focus on hybrid AI-driven sensors, IoT integration, advanced data augmentation, QuEChERS-assisted preprocessing, and explainable AI to improve real-world applicability and interpretability.
Trace contaminants in food pose substantial health risks, yet their spectroscopic detection remains challenging because low analyte abundance and complex matrices cause signal suppression, background interference, and poor quantitative robustness. This structured narrative review synthesizes peer-reviewed studies published from January 2019 to June 2026 and retrieved from Web of Science, PubMed, and ScienceDirect on integrated detection strategies based on surface-enhanced Raman spectroscopy (SERS) and fluorescence spectroscopy (FS). Integrated enhancement is defined as the functional coupling of recognition, enrichment, matrix cleanup, signal transduction, optical amplification, and data interpretation within an analytical workflow. Based on the functional relationships among these modules, strategies are classified as sequential, parallel, or hybrid, while artificial intelligence (AI)-assisted interpretation is treated as a cross-cutting data-analysis layer rather than an independent enhancement mode. Applications involving pesticide residues, mycotoxins and biogenic amines, veterinary drug residues, illicit additives, adulterants and banned residues, and heavy metals are compared in terms of matrix-interference control, analytical robustness, validation depth, portability, and regulatory relevance. Sequential enhancement is most suitable when cleanup, enrichment, indirect transduction, or staged amplification is required; parallel enhancement favors rapid, ratiometric, multichannel, and portable detection; and hybrid enhancement combines upstream purification with downstream multimodal signal generation. AI-assisted interpretation is particularly valuable when spectral overlap, nonlinear responses, or matrix-dependent backgrounds limit direct analysis. Current studies remain constrained by inconsistent validation, overreliance on detection limits, limited external testing of AI models, and inadequate standardization of substrate and probe reproducibility. This workflow-oriented framework supports the rational selection and evaluation of integrated spectroscopic strategies.
Yanyun Wang, Jie Hao, Long Yang et al.· Environmental Research· 0 citations
Routine monitoring of cadmium (Cd) contamination in rice is essential for public health protection and agricultural trade security. Conventional chemical detection methods are environmentally unfriendly, labor-intensive, and slow. This study presents a rapid, accurate classification approach based on near-infrared reflectance spectroscopy (NIRS) for discriminating Cd-contaminated rice from uncontaminated rice. Five spectral preprocessing methods and three variable selection algorithms were systematically evaluated for their influence on model performance. Classification models were developed using partial least squares discriminant analysis (PLS-DA), K-nearest neighbors (KNN), and support vector machines (SVM). Second derivative (2D) preprocessing yielded the greatest performance gains, raising KNN and SVM test-set accuracy from 73% and 88% to 93% and 91%, respectively. Among the variable selection strategies, the successive projections algorithm (SPA) proved most effective. Under optimized conditions, PLS-DA achieved the best overall performance, attaining 92% accuracy, 89% specificity, and 95% sensitivity on the test set. These results demonstrate the strong potential of NIRS coupled with machine learning for rapid, large-scale Cd surveillance in rice, providing robust technical support for grain quality monitoring and low-cadmium variety breeding programs.
Xuexue Miao, Ying Miao, Ni Li et al.· Foods· 0 citations
Foodborne illnesses pose a serious threat to food safety and cause substantial economic losses. Hyperspectral imaging (HSI) has emerged as a powerful tool for rapid and non-destructive identification of foodborne pathogens. However, the high chemical similarity among different pathogen categories presents a challenge for accurate discrimination. To address this issue, we developed an optimized machine learning framework integrated with HSI that incorporates multimodal learning and a multi-head attention mechanism, enabling deeper extraction and fusion of spectral profiles and intensity images features. In visualization analyses, the deeply fused features demonstrated excellent inter-species separability, and the proposed multimodal multi-head attention fusion (MMAF) strategy achieved a high identification accuracy of 96.29%, representing an improvement of 4.17% over the single-modal approach. These results indicate that the optimized HSI approach, driven by advanced machine learning, holds great potential as an effective tool for rapid detection of microbial contamination in food products.
To address the challenge of mixed contamination of foodborne pathogenic bacteria in food, in this study, a machine learning (ML) assisted surface-enhanced Raman scattering (SERS) sensing platform was proposed for multiple and rapid screening of foodborne pathogenic bacteria. 4-Mercaptophenylboronic acid-functionalized gold nanoparticles (AuNPs@4-MPBA) was introduced as SERS substrate and the molecular recognition and electromagnetic enhancement mechanisms for target analytes were elucidated by theoretical simulation. The sensing platform achieved efficient identification and differentiation of single and mixed contaminations of Escherichia coli, Salmonella typhimurium, Shigella, Listeria monocytogenes, and Staphylococcus aureus in three tea samples. The ability of the model to generalize across varying conditions was evaluated using a mixed spectral dataset from different tea sample matrices. After spectral preprocessing, the Random Forest (RF) model was used to classify 31 samples, achieving a classification accuracy of 97.34% and an out-of-bag accuracy of 95.98%, demonstrating excellent stability and generalization capability. The proposed approach enabled the rapid and accurate classification of foodborne pathogenic bacteria in complex food matrices, offering a promising strategy for rapid food safety emergency screening.
Simin Dai, Ceping Yin, Xuejing Fan et al.· Food Quality and Safety· 0 citations
Food safety remains a major concern in global public health and regulatory governance. The illicit adulteration of melamine in food products has raised persistent concerns in food safety due to its potential health risks and the limitations of conventional protein determination methods. This review critically summarizes recent advances in melamine detection in food matrices, covering its physicochemical properties, recognition mechanisms, and major strategies, including conventional physicochemical analysis, optical and spectroscopic detection, electrochemical sensing, molecular recognition, and intelligent integrated platforms. Conventional chromatographic and mass spectrometric methods remain the gold standard for regulatory confirmation, but require costly instrumentation and complex pretreatment. Emerging optical, electrochemical, molecular recognition, and intelligent systems offer rapid, portable, and high-throughput screening, yet still face matrix interference, limited stability, reproducibility issues, insufficient standardization, and challenges in large-scale implementation. Future research should improve robustness, selectivity, portability, data-processing capability, and validation standards. Overall, this review provides a critical assessment of current melamine detection technologies and offers insights into developing practical, reliable, and intelligent platforms for food safety monitoring.
Qianru Meng, Feijie Wang, Shiqiang Ouyang et al.· Critical reviews in analytic...· 1 citation
A new method, called CW-Net, translates the reasoning process of an autonomous vehicle’s AI system into understandable concepts that explain its behavior.
