A neural network prediction model for a psychiatric application

This paper presents a unique application of artificial neural networks used to predict the successful or unsuccessful completion of special education programming for students diagnosed with serious emotional disturbance (SED). In this study, as is common in medical applications, there is an insufficient amount of input data for training and testing the neural network. Bootstrapping and noisy replication of the input data are two techniques used to attempt to compensate for this small amount of available data. While the results would have benefited if more data were available, the results show some promise in being able to correctly predict the successful or unsuccessful completion of SED programming with artificial neural networks, particularly as a diagnostic test.