PANDAS: Paediatric Attention-Deficit/Hyperactivity Disorder Application Software.

Area of Science:

• Neuroscience
• Computational Psychiatry
• Pediatric Psychology

Background:

• Attention-deficit/hyperactivity disorder (ADHD) is a prevalent neuropsychiatric condition affecting individuals across all age groups.
• In South Africa, ADHD affects an estimated 10% of the youth population, necessitating improved diagnostic and management strategies.
• Objective diagnostic tools are crucial for effective ADHD intervention.

Purpose of the Study:

• To develop a novel, quantitative method for aiding in the screening of ADHD.
• To utilize a tablet-based game to collect real-time user data for ADHD classification.
• To train a machine learning model to differentiate between individuals with and without ADHD.

Main Methods:

• A beta-testing phase involved 30 children (ages 5-16), including 11 diagnosed with ADHD.
• A tablet-based game collected gameplay data, from which a feature set was extracted.
• A linear binary support vector machine (SVM) was trained using Sequential Forward Selection (SFS) for feature optimization.

Main Results:

• The trained SVM achieved an overall classification accuracy of 86.5%.
• Test set accuracy reached 85.7%, with a leave-one-out cross-validation (LOOCV) accuracy of 83.5%.
• The model demonstrated a sensitivity of 75%, indicating moderate performance.

Conclusions:

• The study suggests that a quantitative tool for ADHD screening can be developed using gameplay data.
• The SVM model's performance is indicative of the potential for objective ADHD assessment.
• Further research with larger sample sizes is recommended to validate these suggestive findings.