Explainable Artificial Intelligence for Predicting Attention Deficit Hyperactivity Disorder in Children and Adults

Area of Science:

• Neuroscience
• Psychiatry
• Data Science

Background:

• Attention Deficit Hyperactivity Disorder (ADHD) impacts childhood and adult lives, often co-occurring with anxiety, depression, and bipolar disorder.
• These comorbidities can persist into adulthood, necessitating a deeper understanding of their impact and therapeutic implications.

Purpose of the Study:

• To predict ADHD in both children and adults.
• To identify key factors influencing ADHD in different age groups.
• To explore the relationship between ADHD and its common comorbidities.

Main Methods:

• Utilized Machine Learning (ML) and Explainable Artificial Intelligence (XAI) for prediction.
• Employed the National Survey for Children's Health (NSCH) dataset for children (2022).
• Used the ADHD|Mental Health dataset for adults.

Main Results:

• Logistic Regression achieved 99% accuracy for predicting ADHD in children.
• XGBoost demonstrated 100% accuracy for predicting ADHD in adults.
• Identified lack of sleep and excessive smiling/laughing as factors in childhood ADHD, while anxiety and depression are significant for adults.

Conclusions:

• Machine learning models can effectively predict ADHD in children and adults.
• Specific factors like sleep, anxiety, and depression play distinct roles in ADHD across age groups.
• Understanding these factors is crucial for targeted interventions and managing ADHD and its comorbidities.