O' Sullivan Roisin, Cama-Moncunill Raquel, Salter-Townshend Michael, Schmidt Olaf, Monahan Frank J
UCD School of Agriculture and Food Science, University College Dublin, Belfield, Dublin D04 V1W8, Ireland.
UCD School of Mathematics and Statistics, University College Dublin, Belfield, Dublin D04 V1W8, Ireland.
Food Chem X. 2023 Sep 3;19:100858. doi: 10.1016/j.fochx.2023.100858. eCollection 2023 Oct 30.
Scientifically underpinning geographic origin claims will improve consumer trust in food labels. Stable isotope ratio analysis (SIRA) is an analytical technique that supports origin verification of food products based on naturally occurring differences in isotopic compositions. SIRA of five relevant elements (C, H, N, O, S) was conducted on casein isolated from butter (n = 60), cheese (n = 96), and whole milk powder (WMP) (n = 41). Samples were divided into four geographic regions based on their commercial origin: Ireland (n = 79), Europe (n = 67), Australasia (n = 29) and USA (n = 22). A random forest machine learning model built using δC, δH, δN, δO and δS values of all products (n = 197) accurately (88% model accuracy rate) predicted the region of origin with class accuracy of 95% for Irish, 84% for European, 71% for Australasia, and 94% for US products.
从科学角度支持地理原产地声明将提高消费者对食品标签的信任度。稳定同位素比率分析（SIRA）是一种分析技术，可根据同位素组成的自然差异来支持食品的原产地验证。对从黄油（n = 60）、奶酪（n = 96）和全脂奶粉（WMP）（n = 41）中分离出的酪蛋白进行了五种相关元素（C、H、N、O、S）的SIRA分析。根据商业来源，样本被分为四个地理区域：爱尔兰（n = 79）、欧洲（n = 67）、澳大拉西亚（n = 29）和美国（n = 22）。使用所有产品（n = 197）的δC、δH、δN、δO和δS值构建的随机森林机器学习模型准确地（模型准确率为88%）预测了原产地，爱尔兰产品的分类准确率为95%，欧洲产品为84%，澳大拉西亚产品为71%，美国产品为94%。
