Abstarct: Province due to the global belt Digestive diseases among high-risk areas of the world. Therefore, Tehran University of Medical Sciences Research Center of Gastroenterology, the design of various searches to predict the severity of liver damage in patients with fatty liver baxsed on demographic, clinical and laboratory data in different groups of patients suggests. . In this project, the data collected will determine the outcomes of 1464 patients. The neural network used in this research, a neural network MLP and Radial Basis of Neural Networks. Because of Perceptron Perceptron neural network is constructed baxsed on the arithmetic unit. Similar behavior radial basis function networks, biological networks and network-related brain is a multi-laxyer feed forward network with an input laxyer and one of the most powerful neural network learning algorithm, which is formed. Multi-laxyer perceptron networks Pyshkhvrd RBF network compared to the simpler structure is composed of three laxyers fixed: data entry input laxyer, network laxyer, the middle laxyer and output laxyer of RBF is also known as a linear combination of the outputs of all the laxyers between. Given all the training and testing the neural network RBF and MLP, it is concluded that this thesis is an ideal RBF neural network.