SBIR Phase I: Artificial intelligence powered optical spectrometer technology for farm-level milk testing

这个小企业创新研究第一阶段项目致力于改善对牛的护理。乳房炎是一种常见的乳房感染，如果不及早发现，会影响牛奶的质量和数量，并给养殖户造成严重损失。因此，乳制品行业每年损失320亿美元，尽管乳房炎是可以治疗的。该项目实现了农场级牛奶质量检测。患有乳房炎的奶牛会产生可以在牛奶中测量的细胞指标，以及脂肪和蛋白质含量。拟议的技术在挤奶时测量这些数量。乳制品是许多国家的主要营养来源，频繁的检测增加了收入，并通过早期检测最大限度地减少了与乳房炎和抗生素使用有关的成本。该项目集成了移动光学光谱仪技术和软件分析，以测试奶牛场的牛奶质量指标，包括脂肪、蛋白质和体细胞计数。设想的解决方案是一种便携式设备和集成到挤奶系统中的串联单元，以在每头奶牛的水平上指示牛奶质量。软件由云后端支持的移动应用程序构成。应用程序提供测试结果，而云界面生成个人和羊群级别的分析。该项目致力于：使用紫外可见移动光谱对牛奶脂肪、蛋白质和体细胞计数进行高精度(95%)的预测；预测感染的发生和影响的人工智能框架；以及使用近红外光谱检测牛奶成分，如乳糖。将与合作农场和实验室一起收集大量培训数据。现场将安装在线系统，以实现牛奶样本采集和检测的自动化。在获得足够的数据后，将开发一个纵向分析框架来预测长期的羊群表现。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

This Small Business Innovation Research Phase I project addresses improved care of cattle. Mastitis is a common udder infection that impacts both the quality and quantity of milk and causes severe losses for farmers if not detected early. As a result, the dairy industry loses $32 billion each year, even though mastitis is treatable. This project enables farm-level milk quality testing. Cows with mastitis produce cellular indicators that can be measured in milk, along with fat and protein content. The proposed technology measures these quantities at the point of milking. Dairy products are a major source of nutrition in many countries, and frequent testing improves revenues and minimizes both the costs related to mastitis as well as the use of antibiotics through early detection. This project integrates mobile optical spectrometer technology and software analytics to test milk quality indicators, including fat, protein, and somatic cell counts, at the dairy farm. The envisioned solution is a portable device and an in-line unit integrated into milking systems to indicate milk quality at the level of the individual cow. A mobile application supported by a cloud back-end constitutes the software. The app provides test results, whereas the cloud interface generates individual and herd-level analytics. This project addresses: high prediction accuracy (95%) of milk fat, protein and somatic cell counts using ultraviolet-visible mobile spectroscopy; an artificial intelligence framework to predict the occurrence and impact of infections; and detection of milk components, such as lactose, using near-infrared spectroscopy. A large volume of training data will be collected with partner farms and laboratories. In-line systems will be installed on site to automate milk sample collection and testing. After sufficient data are obtained, a longitudinal analytics framework will be developed to predict long-term herd performance.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.