EEG-Based Emotion Recognition Using DWT and Artificial Neural Network: A Case Study on Autism Spectrum Disorder

Autism Spectrum Disorder (ASD) impacts brain development, leading to social communication challenges and interaction. Researchers are increasingly exploring using Artificial Intelligence (AI) to diagnose ASD, interpret their emotions, and search for effective change interventions. This study investigates computer-aided ASD emotion recognition using electroencephalography (EEG) signals. The proposed method implements a four-level Discrete Wavelet Transform (DWT) for feature extraction and an Artificial Neural Network (ANN) to classify three dimensions of emotions: valence, arousal, and dominance. The model achieved 83% accuracy for valence and 96% for arousal and dominance. These findings hold potential for developing an adaptable closed-loop ASD intervention system. In conclusion, EEG-based emotion recognition using DWT and ANN appears promising for identifying emotional challenges in autism. However, further research is needed, considering limitations like sample size and static stimuli.