León Ecay, SaraLópez Maestresalas, AinaraGoenaga Uceda, IrantzuMendizábal Aizpuru, José AntonioInsausti Barrenetxea, Kizkitza2025-08-132025-08-132025León-Ecay, S., López-Maestresalas, A., Goenaga, I., Mendizabal, J. A., Insausti, K. (2025) Authentication of cattle finishing diets (conventional vs. vegetable by-products) using near-infrared spectroscopy. In Katsoulas N. (Ed.), Agricultural Engineerring challenges in existing and new agroecosystems: 1-4 July 2024, Agricultural University of Athens, Greece. (pp. 1159-1163). Hellenic Society of Agricultural Engineers. 978-618-82194-1-0978-618-82194-1-0https://academica-e.unavarra.es/handle/2454/54699This study explored the potential of a portable near-infrared (NIR) spectrophotometer (1200 - 2200 nm; 2-nm bandwidth) to discriminate meat and carcasses from cattle fed two different diets: a control of 90% barley and 10% straw (C, n = 12) and, a ration including 37.5% of vegetable by-products (VBPR, n = 11). At 24 h postmortem spectra were collected on the exterior surface of the left carcass, first, between the 5th and 6th ribs, and afterwards, on the 12th and 13th ribs. After fabrication and 7 d of aging, spectral data was acquired from the intact steaks while keeping muscle integrity. When spectra were collected from the carcasses, partial least squares-discriminant analysis (PLS-DA) correctly classified (%CC) >66.67% in both Train and cross-validation (CV) whereas radial basis function-support vector machine (RBF-SVM) discriminated 100 - 83.33% in Train and CV, respectively, using the full spectrum. Reducing the initial matrix ((λ=501) by interval-PLS (iPLS) led into a >75% of well-sorted carcasses by finishing diet while RBFSVM increased the %CC up to >83.33%. Using both discriminant approaches, carcasses were authenticated with a subtle improvement over intact meat (>90% vs. >75% in Train and >80% vs. >65% in CV for C and VBPR, respectively). Variable importance in projection (VIP) scores showed how variables >1592 nm had higher weight in the discrimination process. The results achieved showed the potential of NIR technology as a sustainable, fast and chemical-free tool to assist the integration of the meat industry into the digital age of connectivity and digitization.application/pdfengCon permiso de la editorial.BeefFeeding dietChemometricsMachine learninginfo:eu-repo/semantics/conferenceObject2025-08-13info:eu-repo/semantics/openAccess