This study investigates the relationship between lameness, milk composition, and rumination behaviour in dairy cows by leveraging sensor-based data for automated monitoring. Lameness was found to significantly impact both rumination and milk production. Lameness was assessed in 24 multiparous Holstein dairy cows throughout early lactation (up to 100 days postpartum), utilising a 1-to-5 scale. Lameness was found to significantly impact both rumination and milk production. On the day of diagnosis, rumination time decreased by 26.64% compared to the pre-diagnosis period (p < 0.01) and by 26.06% compared to healthy cows, indicating the potential of rumination as an early health indicator. The milk yield on the day of diagnosis was 28.10% lower compared to pre-diagnosis levels (p < 0.01) and 40.46% lower than healthy cows (p < 0.05). These findings suggest that lameness manifests prior to clinical signs, affecting productivity and welfare. Milk composition was also influenced, with lame cows exhibiting altered fat (+0.68%, p < 0.05) and lactose (−2.15%, p < 0.05) content compared to healthy cows. Positive correlations were identified between rumination time and milk yield (r = 0.491, p < 0.001), while negative correlations were observed between milk yield and milk fat, protein, and the fat-to-protein ratio (p < 0.001). Additionally, lameness was associated with elevated somatic cell counts in the milk, although sample size limitations necessitate further validation. This study highlights the critical role of rumination and milk performance metrics in identifying subclinical lameness, emphasising the utility of automated systems in advancing dairy cow welfare and productivity. The findings underscore the importance of early detection and management strategies to mitigate the economic and welfare impacts of lameness in dairy farming.

Lameness remains a significant welfare and productivity challenge in dairy farming, often underdiagnosed due to the limitations of conventional detection methods. Unlike most previous approaches to lameness detection that rely on a single-sensor or gait-based measurement, this study integrates four complementary data domains—behavioural, physiological, biochemical, and milk composition parameters—collected from 272 dairy cows during early lactation to enhance diagnostic accuracy and biological interpretability. The main objective of this study was to evaluate and compare the diagnostic classification performance of multiple machine learning (ML) algorithms trained on multimodal data collected at the time of clinical lameness diagnosis during early lactation, and to identify the most influential physiological and biochemical traits contributing to classification accuracy. Specifically, six algorithms—random forest (RF), neural network (NN), Ensemble, support vector machine (SVM), k-nearest neighbors (KNN), and logistic regression (LR)—were assessed. The input dataset integrated physiological parameters (e.g., water intake, body temperature), behavioural indicators (rumination time, activity), blood biochemical biomarkers (non-esterified fatty acids (NEFA), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), gamma-glutamyl transferase (GGT)), and milk quality traits (fat, protein, lactose, temperature). Among all models, RF achieved the highest validation accuracy (97.04%), perfect validation specificity (100%), and the highest normalized Matthews correlation coefficient (nMCC = 0.94), as determined through Monte Carlo cross-validation on independent validation sets. Lame cows showed significantly elevated NEFA and body temperatures, reflecting enhanced lipid mobilization and inflammatory stress, alongside reduced water intake, milk protein, and lactose content, indicative of systemic energy imbalance and impaired mammary function. These physiological and biochemical deviations emphasize the multifactorial nature of lameness. Linear models like LR underperformed, likely due to their inability to capture the non-linear and interactive relationships among physiological, biochemical, and milk composition features, which were better represented by tree-based and neural models. Overall, the study demonstrates that combining sensor data with blood biomarkers and milk traits using advanced ML models provides a powerful, objective tool for the clinical classification of lameness, offering practical applications for precision livestock management by supporting early, data-driven decision-making to improve welfare and productivity on dairy farms.

Subclinical hypocalcemia (SCH) is one of the most prevalent metabolic disorders in early-lactation dairy cows, yet its multifaceted physiological effects are often overlooked due to the absence of clinical symptoms. This study aimed to characterize SCH through an integrative assessment of blood biochemical markers, in-line milk composition, and sensor-derived behavioral traits. Seventy-five Holstein cows between 2 and 21 days in milk were classified into hypocalcemic (group 1) (Ca < 2.0 mmol/L; n = 20) and healthy (group 2) groups (n = 55). Blood samples, milk data, and rumination metrics were evaluated, and group differences were analyzed using Welch’s t-test and Pearson correlations. Cows with SCH exhibited significantly lower concentrations of Ca, PHOS, Mg, ALB, TP, GLUC, and Fe, indicating disruptions in mineral balance, protein metabolism, and energy status. Hepatic indicators (AST, ALT, GGT) did not differ between groups, whereas CREA was significantly lower in hypocalcemic cows, suggesting altered muscle metabolism rather than impaired renal function. Although differences in milk yield, composition, and rumination time did not reach statistical significance, hypocalcemic cows showed consistent biological tendencies toward reduced milk components and lower milk temperature. Correlation analysis revealed strong physiological linkages among Ca, ALB, P, TP, and Fe, underscoring the interconnected nature of mineral and protein metabolism in early lactation. These findings demonstrate that SCH is associated with coordinated biochemical and behavioral changes even in the absence of clinical signs. Integrating blood biomarkers with real-time sensor data provides a more comprehensive understanding of calcium-related metabolic challenges and highlights the potential of precision-livestock technologies for early detection. Future studies incorporating ionized calcium and longitudinal sampling are needed to refine diagnostic thresholds and improve predictive monitoring of SCH in dairy herds.

Metabolic and inflammatory stress during early lactation poses significant risks to dairy cow health and productivity. This study aimed to assess the physiological, metabolic, and inflammatory differences between dairy cows producing low (LL; <4.5%) and high (HL; ≥4.5%) milk lactose, focusing on C-reactive protein (CRP), liver function markers, iron metabolism, and reticulorumen health. A total of 71 clinically healthy lactating multiparous cows (20–30 days postpartum) were monitored using real-time physiological sensors, milk composition analysis, blood biomarkers and continuous reticulorumen pH measurement (every 10 min). Cows in the LL group showed significantly higher aspartate transaminase (AST) activity (p = 0.042), lower serum iron (Fe) concentration (p = 0.013), and reduced reticulorumen pH (p = 0.03). Although CRP concentrations did not differ significantly between groups, correlation analysis revealed positive associations with non-esterified fatty acids (NEFA) (r = 0.335, p = 0.043), reticulorumen pH (r = 0.498, p = 0.002), and body temperature (r = 0.372, p = 0.023). Receiver operating characteristic (ROC) analysis identified gamma-glutamyl transferase (GGT) (AUC = 0.66), AST (AUC = 0.63), and NEFA (AUC = 0.58) as moderate predictors of low milk lactose levels. Conversely, Fe (AUC = 0.66) and reticulorumen pH (AUC = 0.64) showed moderate ability to predict higher lactose content. These results support the integration of milk lactose, liver enzymes, and inflammatory biomarkers into precision health monitoring protocols. The combined use of CRP and milk lactose as complementary biomarkers may enhance the early identification of metabolic stress and support more targeted dairy herd health management.

The objective of the study was to examine the relationship between rumination time and various parameters related to eating and locomotion, including other chews, eating chews, eating time, drinking gulps, bolus counts, chews per minute, activity, and activity change, utilizing RumiWatch technology. The RumiWatch noseband sensor (RWS; ITIN + HOCH GmbH, Feeding Technology, Liestal, Switzerland) was utilized to record time and frequency related to rumination, eating, and movement behaviors. The RumiWatch system (RWS) was put into operation from 1 June 2023 to 30 June 2023. The first two weeks, from 1 June to 14 June 2023 at 7 a.m., served as a period for the cows to acclimate to the RWS, acting as an adjustment phase. Monitoring activities with the RWS commenced on 7 a.m. and lasted until the end of the month, 30 June 2023, with data being recorded daily on an hourly basis. Our findings indicate a significant negative correlation between rumination time and other activity time (r = −0.50), which represents the duration cows allocate to behaviors outside of eating, chewing cud, or distinct movement activities. Additionally, a significant negative correlation was observed between rumination time and eating time (r = −0.54). Furthermore, we observed strong positive correlations with rumination chews (r = 0.84) and bolus (r = 0.75). A weaker positive correlation was found with chews per minute (r = 0.29), while no significant correlation was detected with drinking gulps (r = 0.10). Based on our findings, we recommend the implementation of the RumiWatch System for monitoring rumination and feeding behaviors in lactating dairy cattle. This technology provides valuable insights into cow health and welfare, enabling early detection of potential health issues and improving herd management practices.

This study evaluated six supervised machine learning (ML) models for early detection of subclinical ketosis and low-grade ruminal acidosis in dairy cows during the transition period. Ninety-four Holstein cows were monitored for 21 days postpartum using in-line milk analyzers and intraruminal sensors that continuously recorded milk composition, behavioral, and physiological parameters. Based on clinical examination, blood β-hydroxybutyrate concentration, and fat-to-protein ratio, cows were classified as healthy (n = 44), subclinical ketosis (n = 24), or subclinical acidosis (n = 26). Among the tested models, Random Forest and XGBoost achieved perfect accuracy within this dataset, while Logistic Regression reached 89.5%, Decision Tree 84.2%, and both Naive Bayes and Support Vector Machine 78.9%. These results suggest that ensemble approaches, particularly Random Forest and XGBoost, show strong potential for integration with precision livestock technologies, but their apparent performance should be interpreted cautiously and confirmed in larger, multi-farm studies.