Goi Arianna, Magro Silvia, Lanni Luigi, Boselli Carlo, Marchi Massimo De
Department of Agronomy, Food, Natural Resources, Animals and Environment, University of Padova, Viale dell'Università 16, 35020 Legnaro, Italy.
Istituto Zooprofilattico Sperimentale del Lazio E Della Toscana "M. Aleandri"-National Reference Centre for Ovine and Caprine Milk and Dairy Products Quality, Via Appia Nuova 1411, 00178 Rome, Italy.
Foods. 2025 Jul 7;14(13):2403. doi: 10.3390/foods14132403.
The assessment of milk coagulation properties (MCPs) is crucial for enhancing goat cheese production and quality. In this study, 501 bulk goat milk samples were collected from various farms to evaluate the MCPs. Traditionally, cheesemaking aptitude is evaluated using lactodynamographic analysis, a reliable but time-consuming laboratory method. Mid-infrared spectroscopy (MIRS) offers a promising alternative for the large-scale prediction of goat milk's technological traits. Reference MCP measurements were paired with mid-infrared spectra, and prediction models were developed using partial least squares regression, with accuracy evaluated through cross- and external validation. The ability of MIRS to classify milk samples by coagulation aptitude was evaluated using partial least squares discriminant analysis. Only the model for rennet coagulation time obtained sufficient accuracy to be applied for screening (R = 0.68; R = 0.66; RPD = 2.05). Lower performance was observed for curd-firming time (R = 0.33; R = 0.27; RPD = 1.42) and curd firmness (R = 0.55; R = 0.43; RPD = 1.35). Classification of high coagulation aptitude achieved balanced accuracy values of 0.81 (calibration) and 0.74 (validation). With further model refinement and larger calibration datasets, MIRS may become a resource for the dairy-goat sector to monitor and improve milk suitability for cheesemaking.
乳凝结特性（MCPs）的评估对于提高山羊奶酪的产量和质量至关重要。在本研究中，从各个农场收集了501份散装山羊奶样品，以评估MCPs。传统上，使用乳动力学分析来评估奶酪制作能力，这是一种可靠但耗时的实验室方法。中红外光谱（MIRS）为大规模预测山羊奶的技术特性提供了一种有前景的替代方法。将参考MCP测量值与中红外光谱配对，并使用偏最小二乘回归开发预测模型，通过交叉验证和外部验证评估准确性。使用偏最小二乘判别分析评估MIRS按凝结能力对奶样进行分类的能力。只有凝乳酶凝结时间模型获得了足够的准确性以用于筛选（R = 0.68；R = 0.66；RPD = 2.05）。对于凝乳形成时间（R = 0.33；R = 0.27；RPD = 1.42）和凝乳硬度（R = 0.55；R = 0.43；RPD = 1.35）观察到较低的性能。高凝结能力的分类在校准中达到了平衡准确率值0.81，在验证中达到了0.74。随着模型的进一步优化和更大的校准数据集，MIRS可能成为奶山羊行业监测和改善牛奶用于奶酪制作适宜性的一种资源。
