ADHD-AID: Aiding Tool for Detecting Children's Attention Deficit Hyperactivity Disorder via EEG-Based
1Department of Electronics and Communications Engineering, College of Engineering and Technology, Arab Academy for Science, Technology and Maritime Transport, Alexandria 21937, Egypt.
Biomimetics (Basel, Switzerland)
|March 27, 2024
Summary
This study introduces ADHD-AID, an automated machine learning tool for identifying attention deficit hyperactivity disorder (ADHD) in adolescents. ADHD-AID significantly improves diagnostic accuracy and efficiency, aiding in early intervention.
Area of Science:
- Neuroscience
- Computational Psychiatry
- Biomedical Engineering
Background:
- Attention deficit hyperactivity disorder (ADHD) in adolescents has severe consequences, necessitating early identification and intervention.
- Traditional ADHD diagnostic methods are subjective, time-consuming, and prone to limitations.
- Existing machine learning (ML) models for ADHD detection often use limited features and do not optimize electrode placement or feature selection.
Purpose of the Study:
- To develop and evaluate an automated ML-based tool, ADHD-AID, for accurate and efficient ADHD identification in adolescents.
- To investigate optimal EEG electrode placements and feature selection methods for enhancing ADHD detection accuracy.
- To overcome the limitations of traditional diagnostic techniques and existing ML models.
Main Methods:
- Utilized multi-resolution analysis techniques: variational mode decomposition, discrete wavelet transform, and empirical wavelet decomposition.
- Extracted thirty diverse features (nonlinear, band-power, entropy-based, statistical) from time and time-frequency domains.
- Employed feature selection methods and analyzed EEG electrode placement for optimal ADHD identification.
Main Results:
- ADHD-AID achieved high performance metrics: 0.991 accuracy, 0.989 sensitivity, 0.992 specificity, 0.989 F1-score, and 0.982 Matthews correlation coefficient.
- Demonstrated superior performance compared to previous studies in adolescent ADHD detection.
- Achieved an Area Under the Curve (AUC) of 0.9958 with 10-fold cross-validation.
Conclusions:
- ADHD-AID offers a highly accurate and efficient automated solution for adolescent ADHD identification.
- The tool's performance supports its use as a valuable assistant for clinicians in early ADHD diagnosis.
- Optimized feature extraction, electrode placement, and feature selection significantly enhance diagnostic capabilities.
