The Hybrid Deep Learning Model for Identification of Attention-Deficit/Hyperactivity Disorder Using EEG.

Area of Science:

• Neuroscience
• Artificial Intelligence
• Medical Diagnostics

Background:

• Attention-deficit/hyperactivity disorder (ADHD) diagnosis is challenging due to subjective methods.
• Current diagnostic tools for ADHD lack reliability and timeliness.
• Existing deep learning models like CNNs struggle with temporal data in ADHD diagnosis.

Purpose of the Study:

• To develop a novel hybrid deep learning model for improved ADHD diagnosis.
• To integrate spatial feature extraction and temporal dependency learning from EEG data.
• To enhance the accuracy and reliability of ADHD detection using electroencephalography.

Main Methods:

• A hybrid deep learning model combining Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) was designed.
• The model simultaneously extracts spatial features and learns long-term dependencies from EEG signals.
• The model's performance was evaluated on two public EEG datasets (ADHD and FOCUS).

Main Results:

• The hybrid CNN-LSTM model achieved high classification accuracy: 98.86% on the ADHD dataset and 98.28% on the FOCUS dataset.
• The proposed model demonstrated superior performance compared to state-of-the-art methods for ADHD diagnosis.
• The model effectively captured both spatial patterns and temporal dynamics in EEG data.

Conclusions:

• The hybrid CNN-LSTM model offers a significant advancement in the objective diagnosis of ADHD.
• This deep learning approach shows potential for assisting clinicians in the early and accurate diagnosis of ADHD.
• The model's ability to analyze EEG data provides a more reliable diagnostic tool for attention-deficit/hyperactivity disorder.