Online detection of dairy cow subclinical mastitis using electrical conductivity indices of milk

Abstract

Existing commercial cow mastitis detectors require bulky historical data which may be unavailable or are considered expensive in conventional or small parlours. Thus, the objective of this study is to develop a simple, but without significant sacrifice of accuracy, online cow subclinical mastitis detector for conventional and small parlours. The detective indices are derived merely from the electrical conductivity (EC) of milk using linear discriminant and step regression analyses. The detector was validated on 192 milkings of 48 dairy cows from conventional, small parlours. It had a specificity of 83.7% for healthy quarters, a sensitivity of 46.2% for infected quarters, a prediction accuracy of 90.8% for healthy quarters, and a prediction accuracy of 30.7% for infected quarters. The performance is poorer than commercial detectors, but it is good enough for the dairy industry. This study gives the possibility to give alerts in the milking parlour and no need for animal identification.