Clinical mastitis (CM) needs an effective treatment to eliminate the infection that causes it. CM can be caused by a wide variety of pathogens. Knowing the causal pathogens and, subsequently, using appropriate treatments, will increase the cure rate of CM. Bacteriological culturing will provide information about the causal pathogens. However, because CM needs a treatment immediately after detection, culture information comes too late. In the absence of bacteriological culture results, several other sources of information are available on a dairy farm that could aid in the diagnosis of the causal pathogens of a CM case. In previous studies, various classification models for pathogen identification for CM were constructed. However, the predictive performance of these models varied strongly. A further disadvantage of these models was that they only returned the most likely causal pathogen to a farmer. For choosing among treatment options, a probability distribution for the causal pathogens of a CM case would be more informative as it reveals the uncertainty involved in the classification. For instance, almost equal probabilities for two or more causal pathogens would support the decision for a broad spectrum antibiotic treatment while a very high probability for a particular pathogen would support the choice for a more specific treatment.
展开▼