Automatic Diagnosis of Attention Deficit Hyperactivity Disorder with Continuous Wavelet Transform and Convolutional...

Area of Science:

• Neuroscience
• Psychiatry
• Artificial Intelligence

Background:

• Attention deficit hyperactivity disorder (ADHD) significantly impacts a child's education and social development.
• Previous research has established a link between temperament traits and ADHD.
• No prior machine learning studies have focused on diagnosing ADHD using a dataset derived from temperament characteristics.

Purpose of the Study:

• To evaluate the efficacy of a semi-automatic expert decision support system for ADHD diagnosis.
• To investigate the diagnostic potential of temperament characteristics in ADHD.
• To establish a foundation for AI-based ADHD diagnostic systems.

Main Methods:

• Utilized deep learning models for classification.
• Employed Continuous Wavelet Transform to generate signal images from the dataset.
• Developed a semi-automatic expert decision support system.

Main Results:

• The Squeeze Net model demonstrated the highest classification success at 88.33%.
• The study successfully created a dataset based on temperament characteristics for ADHD analysis.
• The findings highlight the potential of temperament traits in ADHD diagnosis.

Conclusions:

• An artificial intelligence-based automatic system for ADHD diagnosis is feasible.
• The study underscores the relationship between temperament characteristics and ADHD diagnosis.
• The developed model offers a novel approach for ADHD assessment.