Aggarwal Ashish, Mishra Akanksha, Tabassum Nazia, Kim Young-Mog, Khan Fazlurrahman
School of Bioengineering and Biosciences, Lovely Professional University, Phagwara 144001, Punjab, India.
Marine Integrated Biomedical Technology Center, The National Key Research Institutes in Universities, Pukyong National University, Busan 48513, Republic of Korea.
Foods. 2024 Oct 21;13(20):3339. doi: 10.3390/foods13203339.
Mycotoxin contamination of foods is a major concern for food safety and public health worldwide. The contamination of agricultural commodities employed by humankind with mycotoxins (toxic secondary metabolites of fungi) is a major risk to the health of the human population. Common methods for mycotoxin detection include chromatographic separation, often combined with mass spectrometry (accurate but time-consuming to prepare the sample and requiring skilled technicians). Artificial intelligence (AI) has been introduced as a new technique for mycotoxin detection in food, providing high credibility and accuracy. This review article provides an overview of recent studies on the use of AI methods for the discovery of mycotoxins in food. The new approach demonstrated that a variety of AI technologies could be correlated. Deep learning models, machine learning algorithms, and neural networks were implemented to analyze elaborate datasets from different analytical platforms. In addition, this review focuses on the advancement of AI to work concomitantly with smart sensing technologies or other non-conventional techniques such as spectroscopy, biosensors, and imaging techniques for rapid and less damaging mycotoxin detection. We question the requirement for large and diverse datasets to train AI models, discuss the standardization of analytical methodologies, and discuss avenues for regulatory approval of AI-based approaches, among other top-of-mind issues in this domain. In addition, this research provides some interesting use cases and real commercial applications where AI has been able to outperform other traditional methods in terms of sensitivity, specificity, and time required. This review aims to provide insights for future directions in AI-enabled mycotoxin detection by incorporating the latest research results and stressing the necessity of multidisciplinary collaboration among food scientists, engineers, and computer scientists. Ultimately, the use of AI could revolutionize systems monitoring mycotoxins, improving food safety and safeguarding global public health.
食品中的霉菌毒素污染是全球食品安全和公众健康的一个主要问题。人类使用的农产品被霉菌毒素（真菌的有毒次级代谢产物）污染，对人类健康构成重大风险。霉菌毒素检测的常用方法包括色谱分离，通常与质谱联用（准确但样品制备耗时且需要技术熟练的技术人员）。人工智能（AI）已被引入作为食品中霉菌毒素检测的新技术，具有高可信度和准确性。这篇综述文章概述了最近关于使用人工智能方法发现食品中霉菌毒素的研究。新方法表明，多种人工智能技术可以相互关联。实施深度学习模型、机器学习算法和神经网络来分析来自不同分析平台的详细数据集。此外，本综述重点关注人工智能与智能传感技术或其他非常规技术（如光谱学、生物传感器和成像技术）协同工作的进展，以实现快速且危害较小的霉菌毒素检测。我们质疑训练人工智能模型所需的大量多样数据集的要求，讨论分析方法的标准化，并讨论基于人工智能的方法获得监管批准的途径，以及该领域其他首要问题。此外，本研究提供了一些有趣的用例和实际商业应用，其中人工智能在灵敏度、特异性和所需时间方面能够优于其他传统方法。本综述旨在通过纳入最新研究成果并强调食品科学家、工程师和计算机科学家之间多学科合作的必要性，为人工智能支持的霉菌毒素检测的未来方向提供见解。最终，人工智能的使用可能会彻底改变霉菌毒素监测系统，提高食品安全并保障全球公众健康。
