Abstarct: Artificial intelligence becomes an important tool for decreasing human rules in many domains. Many tasks were done by people with high cost in the past. But now we can do them with intelligent systems; and in many cases intelligent systems have less cost and more punctuality. But decreasing the role of experts was challenging always. Autism is a wide range of behavioral disorders. The common denominator of patients with this disorder is stereotyped patterns of behavior, difficulties in oral communication and social interaction. This problem causes autistic children can't learn and grow like others. Although there is no certain treatment for autism, but we can help patient to interact environment with some trainings. In this thesis we present a new method for measuring amount of autistic disorders in people. We call this method Moving Picture for Concentration Measurement (MPCM). So, we design a computer game with two group of shapes, circles and squares. Our consideration is on measuring focus, especially when shapes are moving. We asked different children to play with MPCM and registered their performance. This children are in two groups; the first group are diagnosed autistic and the second one are normal children. Normal children help us in comparison. The severity of disorder is presented as a fuzzy number. For autistic children experts determined this number and for normal group we assign a small number. For judging about people performance, we need a way to represent the existing knowledge in this field and find a map from inputs to outputs. The artificial neural networks are a suitable structure for this target. In this thesis we use them. Finally we put every sample in one group of disorder's severity; normal, mild, moderate and severe.