Autism Disorder Diagnosis from EEG Signals with Fuzzy-Deep Belief Neural Network

One of the disorders that occurs in children is autism which severely reduces the ability to communicate with others. High immobility, lack of social activities and excessive mental conflicts are the most important factors of this disorder. Many researchers have tried to diagnose autism disorder in early times, but so far they have not succeeded in providing a precise method with high accuracy. Providing a method that can help doctors in early diagnosis of autism disorder can be used as a decision management system in the field of medical and an assistant for doctors. Artificial intelligence gives researchers the ability to create methods for such a system by studying and carefully examining dimensions and features based on data. The use of EEG signal in the autism diagnosis disorder can be useful, so researchers and doctors have presented a data called BioGS, which contains EEG signals with the characteristics of autism disorder. The application of fuzzy logic in modeling can be interesting due to the uncertainty in the type of feature recognition and its training with the help of a deep learning model, which is a type of deep belief network may improve the results. The simulation of this method was done in MATLAB and the results indicated what factors represent the autism diagnosis and estimated the accuracy rate up to ۹۸.۳۹% and the results were more promising than the previous methods.

Keywords:

Autism Disorder , Medical Intelligent System , Deep Belief Neural Network , Deep Learning , Fuzzy Logic