Application of an electronic nose for the diagnosis of ketosis in dairy cows

Abstract

Ketosis in dairy cows is a nutritional and metabolic disease that severely affects the milk industry and dairy cow health. An electronic nose can identify volatile organic compounds (VOCs). This study applied the electronic nose to detect VOCs in dairy cow milk, feces, and blood. Blood ketone bodies (BHBA) were measured, and 10 ketosis cows, 71 subclinical ketosis cows, and 75 healthy cows were considered the research subjects. Cow milk, feces, and blood samples were collected for electronic nose testing. The measured VOC data were processed through multivariate analysis and machine learning algorithms. The electronic nose analysis results were validated through the GC-MS analysis of the milk samples. The results unveiled spatial differences in the electronic nose VOC values of milk, feces, and blood, and the cumulative contribution rates of both principal component variables (PC) 1 and PC2 reached more than 95%. The machine learning algorithm results revealed that the accuracy of the milk probabilistic neural network (PNN) was 87.5%, the stool support vector machine was 100%, and the blood PNN was 66.7%. This result revealed that electronic nose detection can effectively differentiate healthy, subclinical ketosis, and clinical ketosis cows. According to the milk GC-MS results, significant differences were noted in the VOC content of milk samples between the different groups.The aforementioned studies revealed that electronic nose can aid in the effective prediction and diagnosis of ketosis in dairy cows. This lays the foundation for establishing a new method for diagnosing ketosis in dairy cows.