Beetle Swarm-based Multi Verse Optimization for the Autism Spectrum Disorder Detection via EEG Signal Analysis: A Novel Hybrid Deep Learning Concept

Abstract

Traditional automated EEG-based diagnosis of ASD typically focuses on distinguishing children with autism spectrum disorder (ASD) from their typically developing peers without addressing the severity of their autistic traits. Using EEG research to identify possible variations among children having mild and severe ASD is a serious challenge. Therefore, this paper plans to diagnose ASD through EEG signal processing and analysis. This dataset was initially collected from the ASD dataset. Median filtering and normalization complete the preprocessing. The signal is converted into an image using the STFT method. GLCM is used to extract features from the transformed image. Since the extracted features are very long, PCA is used to select the essential features. The classification uses a hybrid deep learning model based on these ideally picked features. In this case, SVM replaces CNN’s fully connected layer. By integrating BSO and MVO, the novel Optimized BS-MVO improves CNN’s learning rate and the number of iterations of SVM, achieving an improved CNN that increases accuracy as an exercise or objective function. The final output is classified (ASD or not) using this enhanced CNN. The accuracy of the proposed BS-MVO at 90% learning percentage is higher at 94.49% compared to GA with 92.90%, WOA at 89.99%, MVO at 93.91%, and BSO at 91.60%, respectively. The simulation results show that the proposed method outperforms the traditional methods in all aspects.