Development of a new clinical mastitis detection method for automatic milking systems

Area of Science:

• Animal Science
• Veterinary Medicine
• Dairy Science

Background:

• Clinical mastitis (CM) is a significant concern in dairy farming, impacting animal welfare and economic viability.
• Automatic milking systems (AMS) generate vast amounts of electronic data that can be leveraged for disease detection.

Purpose of the Study:

• To investigate the efficacy of using electronic data from AMS support software for accurate clinical mastitis detection.
• To develop and validate a predictive model for CM using various electronic measurements.

Main Methods:

• Logistic mixed models were employed to analyze data from 12 electronic measurements.
• A backward elimination process identified key predictors: milk yield (MY), electrical conductivity (EC), average milk flow rate (MF), incompletely milked quarters (IM), MY per hour (MYH), and EC per hour (ECH).
• Model performance was evaluated using receiver operating characteristic curves and area under the curve (AUC) on independent datasets.

Main Results:

• Six key measurements (EC, ECH, MY, MYH, MF, IM) combined with cow and quarter effects yielded the highest predictive accuracy (90% sensitivity, 91% specificity, AUC 0.96).
• The model demonstrated robust performance across different datasets and could predict CM 1-3 days prior to clinical diagnosis, albeit with reduced accuracy.
• Nine individual measurements showed significant mastitis detection ability.

Conclusions:

• Electronic data from AMS can be effectively utilized to develop accurate predictive models for clinical mastitis.
• Incorporating multiple electronic measurements enhances mastitis detection accuracy, leading to improved on-farm management and alerts.
• Further development of new indices based on these findings is expected to improve the utility and accuracy of mastitis detection systems.