Classification of ADHD and Non-ADHD Subjects Using a Universal Background Model

Area of Science:

• Neuroscience
• Biomedical Engineering
• Pediatric Neurology

Background:

• Attention Deficit Hyperactivity Disorder (ADHD) significantly impacts children, predominantly boys.
• Current ADHD diagnosis relies on subjective observation and interviews, highlighting the need for objective diagnostic tools.
• Existing diagnostic methods lack objective measures, leading to potential inaccuracies and delays in treatment.

Purpose of the Study:

• To develop and evaluate an objective method for ADHD detection using electroencephalography (EEG).
• To adapt techniques from speaker verification (imposter problem) for ADHD identification.
• To assess the efficacy of a novel EEG-based approach for ADHD detection in young children.

Main Methods:

• Utilized multi-channel EEG data from children during an attention network task.
• Employed autoregressive model parameters as features for ADHD detection.
• Applied Gaussian mixture models to create ADHD and universal background models for a likelihood ratio detector.
• Investigated the impact of data contamination on detection performance.

Main Results:

• The proposed EEG-based method achieved high probability of detection simultaneously with a low equal error rate.
• Performance was evaluated using standard metrics like area-under-the-curve and equal-error-probability.
• Results were based on a limited dataset of approximately 6-year-old males.

Conclusions:

• The developed EEG-based approach shows promise as an objective tool for ADHD detection.
• This method offers a potential improvement over subjective diagnostic methods.
• Further validation on diverse datasets is warranted to confirm generalizability.