Attention deficit and hyperactivity from EEG signs determination of disorders by machine learning methods

Attention Deficit Hyperactivity Disorder (ADHD) is a condition commonly observed in children, characterized by symptoms such as attention deficits, hyperactivity, behavioral problems, and difficulties related to education. This study aims to assess the diagnosis of ADHD using electroencephalography (EEG) signals, marking a significant step forward. The research includes EEG data obtained from 30 children diagnosed with ADHD and 30 healthy controls. The EEG data were first processed for noise reduction and then classified using deep learning models such as ConvMixer, ResNet50, and ResNet18. The findings of the study indicate that the ConvMixer model achieves high classification accuracy with low computational resources. This underscores the availability of lighter models for ADHD diagnosis, providing a significant advantage in practical applications. Furthermore, the research on the usability of EEG signals in ADHD diagnosis examined the effects of different channels. Through these analyses, it was determined that the T8 channel is particularly effective in diagnosing ADHD. This provides valuable insight into which channels to focus on in EEG-based ADHD diagnosis. In conclusion, this study demonstrates the feasibility of EEG-based methods for ADHD diagnosis and the effectiveness of deep learning models in analyzing EEG signals. It also sheds light on the availability of lighter models and the importance of certain EEG channels in practical applications.