Ensemble learning of attention-based BiLSTM networks for ADHD detection from EEG signals code

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Early detection of Attention Deficit Hyperactivity Disorder (ADHD) is crucial for effective intervention and improving children's long-term mental health outcomes.
• Current diagnostic methods for ADHD often rely on subjective assessments, leading to inconsistencies and potential delays in diagnosis.
• Developing objective and reliable diagnostic tools is essential for timely and accurate ADHD identification.

Purpose of the Study:

• To present an advanced ensemble learning model for the early and objective detection of ADHD in children using electroencephalogram (EEG) data.
• To enhance feature extraction and classification accuracy by integrating multiple data processing techniques with a deep learning architecture.
• To demonstrate the efficacy of the proposed model in achieving high diagnostic accuracy for ADHD.

Main Methods:

• An ensemble learning approach was implemented, featuring a Parallel Attention-Based BiLSTM (PABiLSTM) model.
• EEG data was processed to extract features using spectrograms, fractal dimensions, and recurrence plots.
• ResNet-50 was utilized to enhance feature extraction and improve the classification performance of the PABiLSTM model.

Main Results:

• The proposed PABiLSTM model, combined with ResNet-50 processed features, achieved high diagnostic accuracy.
• Experimental results on two independent datasets demonstrated classification accuracies of 98.91% and 99.10%.
• The findings indicate a significant improvement in diagnostic accuracy compared to traditional subjective methods.

Conclusions:

• Deep learning models applied to EEG data offer a highly accurate and objective method for ADHD diagnosis in children.
• This approach holds significant promise for developing more effective intervention strategies and treatment pathways for ADHD.
• The study highlights the potential of AI in revolutionizing pediatric mental health diagnostics.