Figure 1From: Would artificial neural networks implemented in clinical wards help nephrologists in predicting epoetin responsiveness? Schematic representation of an artificial neural network. A typical ANN consisting of one input layer, two hidden layers and one output layer is represented. The basic structure, fed forwards and trained by back-propagation is called Multilayer Perceptron (MLP) while models designed with connections jumping over hidden layers (---) are called Generalized Feedforward Networks (GFN).Back to article page