Abstract
In this research discuss the concept of appropriate form, by examining mathematical behavior for three models represent neural networks are (GRNN, BPNN, PNN), were applied two types of medical data are (osteoporosis and weaknesses auditory) and different in the way of classification and spaces Input and output, and show through the application of these data and suitability models with neural networks in terms of the Domain and Range the network (PNN) is the best in the diagnosis of audio data through average MSE, and network (GRNN) is better diagnose bone crisp data (which are more complex) and the network (BPNN) is the most generalization, especially when test data are large compared with the training data.
