Development of a Decision Support Model for Screening Attention-deficit Hyperactivity Disorder with Actigraph-based...

Area of Science:

• Pediatric neurology
• Biomedical engineering
• Behavioral science

Background:

• Traditional Attention Deficit Hyperactivity Disorder (ADHD) screening relies on subjective questionnaires.
• Existing objective ADHD assessment methods lack widespread applicability.
• There is a need for objective, accurate, and accessible ADHD screening tools.

Purpose of the Study:

• To develop a decision support model for ADHD screening.
• To utilize a 3-axial actigraph to monitor children's school activities objectively.
• To establish an accurate and efficient ADHD screening method.

Main Methods:

• 153 children wore actigraphs on their non-dominant wrists during school.
• Activity features (mean, variance, low/high-level activity ratios) were extracted.
• Two decision-tree models (C5.0 algorithm) were built using data from general school hours and class time only.

Main Results:

• Model A (whole hours) achieved 99.30% accuracy, 1.0000 sensitivity, and 0.992 specificity.
• Model B (class time) achieved 98.59% accuracy and 0.985 specificity.
• Both models demonstrated high predictive values and area under the curve (AUC).

Conclusions:

• Objective ADHD screening is feasible using a simple, wearable sensor.
• The proposed model provides accurate ADHD detection with minimal time and cost.
• This approach offers significant public health benefits for early ADHD identification.