Early attention-deficit/hyperactivity disorder (ADHD) with NeuroDCT-ICA and rhinofish optimization (RFO) algorithm...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Biomedical Engineering

Background:

• Existing methods for detecting Attention Deficit Hyperactivity Disorder (ADHD) face limitations in data processing, accuracy, and computational efficiency.
• Deep Learning (DL) frameworks present a transformative potential for enhancing ADHD detection systems.

Purpose of the Study:

• To develop a sophisticated ADHD detection framework that overcomes the drawbacks of current approaches.
• To leverage DL for improved accuracy, data processing, and reduced computational time in ADHD identification.

Main Methods:

• A novel NeuroDCT-ICA module was developed for preprocessing raw EEG data, focusing on noise elimination and feature extraction.
• A new RhinoFish Optimization (RFO) algorithm was introduced for optimal feature selection, enhancing system stability and data processing.
• An ADHD-AttentionNet, a DL-based model, was employed as the core component for ADHD identification.

Main Results:

• The proposed model achieved high performance metrics, including an accuracy of 98.52%, an F-score of 98.26%, and a specificity of 98.16%.
• The NeuroDCT-ICA module effectively eliminated noise and extracted informative features from EEG data.
• The RFO algorithm enhanced system stability and data processing capacity.

Conclusions:

• The developed DL-based framework offers a more accurate and efficient method for ADHD detection.
• The integration of NeuroDCT-ICA, RFO, and ADHD-AttentionNet demonstrates superior performance in identifying ADHD-related patterns.
• This advanced approach holds significant promise for clinical applications in ADHD diagnosis.