Neural Network and Artificial Intelligence Study in Psychiatric Intelligent Diagnosis

We have applied intelligent control theory into medical psychiatric diagnosis. During studying, we find selecting suitable neural network structure is very important to BP network. Suitable neural network structure will bring less error to diagnosis system. It is key to success or failure of psychiatric intelligent diagnosis system. We have found some rules that suit our diagnosis system. Such as : full connecting mode is better and at the same time adding suitable hidden node number can improve convergence effect and reduce error of network. But adding hidden layer number doesn't always improve network convergence effect under our studying. At the same time it makes network convergence speed to become slower and makes network training time to increase. We build an intelligent diagnosis system. Comparing the diagnosis by computer with the senior child psychiatrists, the consistent rate of intelligent diagnosis is 99%.