The use of artificial neural networks to diagnose mastitis in dairy cattle

The use of milk sample categorization for diagnosing mastitis using Kohonen’n Self-Organizing Feature Map (SOFM) is reported. Milk traits data of 14 weeks of milking from commercial dairy cows in New Zealand was used to train and test a SOFM network to discriminate the samples into its appropriate health category. The SOFM was useful in discriminating data patterns into four separate mastitis categories. Also, several Artificial Neural Networks were tested to predict missing data from the recorded milk traits. A multi-layer perceptron (MLP) network proved to be most accurate (R = 0.84, r = 0.93) when compared to other MLP, Elman, Jordan or linear regression methods. It is concluded that the SOFM can be used as a decision tool for the dairy farmer to reduce the incidence of mastitis in the dairy herd.