Designing a clinical decision support system for differentiating attention deficit/hyperactivity disorder from emotional-behavioral disorders with similar symptoms: Comparison of two common artificial neural networks

The incidence of behavioral disorders among children and teenagers has recently found interest among researchers. Due to their harsh and offensive mood, most of the affected individuals are rejected from their families and society, which leads to decreasing their chance of normal development, so it is important to identify and treat them in the course as early as possible. Diagnosing and distinguishing attention deficit/hyperactivity disorder from other similar behavioral disorders such as depression, anxiety, comorbid depression and anxiety, and conduct disorders is one of the most important and essential proceedings in field of child psychology disorders.
The samples consisted of 271 children, including 44 with ADHD, 31 with conduct, 35 with depression, 33 with mixed depression and anxiety, and 41 with anxiety as well as 87 children with normal but temporarily aggressive behavior. In the present study, two different decision support systems, multilayer perceptron and Radial Basis Function neural networks, were designed and compared based on the signs and symptoms.
The mean of accuracy of the networks in diagnosis and distinguishing reached to 95.57 and 96.30 percentages with MLP and RBF, respectively. Therefore, the designed decision support systems, especially RBF, was observed to be a reliable assistant for the experts in the diagnosis and distinguishing the mentioned behavioral disorders.
Both designed systems, especially RBF, can be used as a reliable device for distinguishing, diagnosing, and also screening of child emotional and behavioral disorders