Taheri Salman, Andrade Jelmir Craveiro de, Conte-Junior Carlos Adam
Center for Food Analysis (NAL), Technological Development Support Laboratory (LADETEC), Federal University of Rio de Janeiro (UFRJ), Rio de Janeiro, RJ, Brazil.
Laboratory of Advanced Analysis in Biochemistry and Molecular Biology (LAABBM), Department of Biochemistry, Federal University of Rio de Janeiro (UFRJ), Rio de Janeiro, RJ, Brazil.
Crit Rev Food Sci Nutr. 2024 Dec 2:1-27. doi: 10.1080/10408398.2024.2435597.
This review systematically explores the emerging perspectives on analytical techniques and machine learning applications in food metabolomics, with a focus on their roles in the era of Industry 4.0. The study emphasizes the utilization of chromatography-mass spectrometry and proton nuclear magnetic resonance spectroscopy as primary tools for metabolic profiling, highlighting their respective strengths and limitations. LC-MS, known for its high sensitivity and specificity, faces challenges such as complex data interpretation and the need for advanced computational tools.H NMR offers reproducibility and quantitative accuracy but struggles with lower sensitivity compared to mass spectrometry. The review also delves into the integration of multivariate data analysis techniques like principal component analysis and partial least squares-discriminant analysis, which enhance data dimensionality reduction and pattern recognition, yet require careful validation to avoid overfitting. Furthermore, the application of machine learning algorithms, including random forests and support vector machines, is discussed in the context of improving classification and predictive tasks in food metabolomics. Practical applications of these technologies are demonstrated in areas such as quality control, nutritional studies, and food adulteration detection. The review emphasizes the need for standardization in methodologies and the development of more accessible and cost-effective analytical workflows. Future research directions include enhancing the sensitivity of H NMR, integrating metabolomics with other omics technologies, and fostering data sharing to build comprehensive reference libraries. This review aims to provide a comprehensive and critical overview of the current advancements and future potentials of analytical techniques and machine learning in food metabolomics, aligning with the goals of Industry 4.0.
本综述系统地探讨了食品代谢组学中分析技术和机器学习应用的新观点，重点关注它们在工业4.0时代的作用。该研究强调了将色谱-质谱联用和质子核磁共振光谱作为代谢物谱分析的主要工具，突出了它们各自的优势和局限性。液相色谱-质谱联用（LC-MS）以其高灵敏度和特异性而闻名，但面临着复杂的数据解读以及对先进计算工具的需求等挑战。核磁共振氢谱（1H NMR）具有可重复性和定量准确性，但与质谱相比灵敏度较低。该综述还深入探讨了主成分分析和偏最小二乘判别分析等多元数据分析技术的整合，这些技术可增强数据降维和模式识别能力，但需要仔细验证以避免过拟合。此外，还讨论了包括随机森林和支持向量机在内的机器学习算法在改善食品代谢组学中的分类和预测任务方面的应用。这些技术在质量控制、营养研究和食品掺假检测等领域展示了实际应用。该综述强调了方法标准化的必要性以及开发更易于使用和成本效益更高的分析工作流程的重要性。未来的研究方向包括提高核磁共振氢谱的灵敏度、将代谢组学与其他组学技术整合以及促进数据共享以建立综合参考库。本综述旨在全面且批判性地概述食品代谢组学中分析技术和机器学习的当前进展及未来潜力，与工业4.0的目标保持一致。
