Milk quality has a growing importance for farmers as component-based pricing becomes more widespread. Food quality and precision manufacturing techniques demand consistent milk composition. Udder health, general cow condition, environmental factors, and especially feed composition all influence milk quality. The large volume of routinely collected milk data can be used to build prediction models that estimate valuable constituents from other measured parameters. In this study, casein was chosen as the target variable because of its high economic value. We developed a multiple linear-regression model and a feed-forward neural network model to estimate casein content from twelve commonly recorded milk traits. Evaluated on an independent test set, the regression model achieved R² = 0.86 and RMSE = 0.018%, with mean bias = +0.003% and slope bias = −0.10, whereas the neural network improved performance to R² = 0.924 and RMSE = 0.084%. In silico microgreen inclusion from 0% to 100% of dietary dry matter raised the predicted casein concentration from 2.662% to 3.398%, a relative increase of 27.6%. To extend practical applicability, a simulation module was created to explore how microgreen supplementation might modify milk casein levels, enabling virtual testing of dietary strategies before in vivo trials. Together, the predictive models and the microgreen simulation form a cost-effective, non-invasive decision-support tool that can accelerate diet optimization and improve casein management in precision dairy production.

Precision agriculture brings new artificial intelligence techniques closer to everyday farming. Agriculture historical data is easily available, so using this data to teach a machine-learning model, offers various opportunities to enhance farming efficiency. In our study, we develop a machine learning model to estimate some linear traits of Limousin sires (sore for muscularity, length of the rump, muscularity of breast and muscularity of the width of rump), based on a phenotypic score, using artificial intelligence, in Hungary. Phenotypic scores are usually given by the experts in field. Before scoring, many measurements are made on the animals, which takes time and places a high stress on the cattle. A hands-on prediction application can make the whole process faster, and more comparable, regardless of the expert who created the scoring. We found that after collecting sufficient data from previous observations it is possible to train specifically selected artificial intelligence (AI) algorithms to predict linear traits in Limousin breeding bulls. Machine learning (ML) was used to predict the score values for muscularity, length of the rump, muscularity of the breast and muscularity of the width of the rump for this study. We found no similar experiments for the usage of AI algorithms to predict these variables. The coefficient of determination (R²) of the algorithm, in this study, provided the following range values: (R²=0.77 to 0.86).

Milk quality is an important factor both for the farmers to be able to sell their products and for the milk industry to be able to plan its production based on quantity and quality. Milk quality has a direct link with cow health, more specifically with utter health. One of the most common utter diseases is mastitis. It always captures a lot of interest based on its frequency and cost as a dairy disease which eventually leads to an involuntary and premature culling of milking cows and decreased milk yield. The genetic evaluation of mastitis is very difficult as it is a low heritable trait and categorical in nature [2]. That is why it is necessary to find markers that could predict the occurrence of mastitis. One of the widely used such markers is the somatic cell count (SCC) [9] which is considered to be the most suitable indicator trait for mastitis resistance given its medium to high genetic correlation with mastitis and its greater heritability than mastitis. The SCC is also easy to record in the practice. The selection for lower SCC in milk has a positive effect on the incidence of mastitis. The selection against high SCC also does not deteriorate the immune system of cattle and decreases the risk of infection at the same time. The genetic evaluation [1] of this trait is mostly based on somatic cell score (SCS), a logarithmic transformation of SCC to achieve normality of distribution. In our study, we used the milk database of Holstein cows from 3 different farms. From each farm, we had altogether 8000 samples tested. The samples were analyzed using chemical methods every month for a year. 11 different types of data were recorded from each sample. Our aim was to find the best mixture of recorded data that would predict the value of linearized somatic cell count. After the logarithmic linearization the SCC results were divided into 3 main groups (based on the probability of mastitis). Thus our prediction problem turned into a classification problem. We used machine learning to train our algorithm. We experimented with different types of classification methods and found good results for the prediction of SCC in milk samples. We changed the input variables as not all the 9 measured input variables will be necessary for good prediction results. Our preliminary results show that using machine learning it is possible to build a model that can be used to predict mastitis in dairy cows based on variables generally analyzed during milk quality checking tests.