Scholarlink

<p>Food allergy is a sensitivity to a food or one of its components that triggers the immune system to react. It affects around 8% of children and 10% of adults. This study investigates the use of near-infrared spectroscopy (NIRS) combined with artificial intelligence (AI) approaches for rapid and non-destructive detection of lipid transfer proteins (LTP), a cause of severe food allergy. Unlike traditional chemical detection methods, which are time-consuming, costly, and often require complex sample preparation, NIRS offers a fast, non-invasive alternative suitable for routine food safety screening. NIRS spectra were acquired from a wide variety of foods, both with and without LTP, using a miniature spectrometer. Three deep learning architectures, convolutional neural networks (CNN), vision transformers (ViT), and TabTransformer, were employed to classify foods according to the presence of LTP, with hyperparameters optimized via Bayesian optimization. The findings indicated that ViT and CNN-based models had significant potential, achieving accuracies and F1 scores above 90%. Key wavelengths in the 1325–1455 nm range were identified as useful for identifying foods with LTP, reflecting changes in water, fat, and protein content. Additionally, AI explainability methods (SHAP, LIME, and Grad-CAM) were applied to better understand model decisions. The study demonstrates the potential of NIRS combined with AI as a rapid, reliable, and non-destructive tool for improving food allergy detection and safety.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Detection of LTP in Food Using Near-Infrared Spectroscopy and Explainable Artificial Intelligence

  • 期刊论文

  • 发表日期：

    2025年9月1日

查看全文
  • Ainhoa Osa-Sanchez,
  • Itxasne Del Barrio,
  • Ganeko Bernardo-Seisdedos,
  • Amaia Mendez-Zorrilla,
  • Begonya Garcia-Zapirain

摘要

Food allergy is a sensitivity to a food or one of its components that triggers the immune system to react. It affects around 8% of children and 10% of adults. This study investigates the use of near-infrared spectroscopy (NIRS) combined with artificial intelligence (AI) approaches for rapid and non-destructive detection of lipid transfer proteins (LTP), a cause of severe food allergy. Unlike traditional chemical detection methods, which are time-consuming, costly, and often require complex sample preparation, NIRS offers a fast, non-invasive alternative suitable for routine food safety screening. NIRS spectra were acquired from a wide variety of foods, both with and without LTP, using a miniature spectrometer. Three deep learning architectures, convolutional neural networks (CNN), vision transformers (ViT), and TabTransformer, were employed to classify foods according to the presence of LTP, with hyperparameters optimized via Bayesian optimization. The findings indicated that ViT and CNN-based models had significant potential, achieving accuracies and F1 scores above 90%. Key wavelengths in the 1325–1455 nm range were identified as useful for identifying foods with LTP, reflecting changes in water, fat, and protein content. Additionally, AI explainability methods (SHAP, LIME, and Grad-CAM) were applied to better understand model decisions. The study demonstrates the potential of NIRS combined with AI as a rapid, reliable, and non-destructive tool for improving food allergy detection and safety.