Fusion of Multi-Task Neurophysiological Data to Enhance the Detection of Attention- Deficit/Hyperactivity Disorder

Area of Science:

• Neuroscience
• Developmental Psychology
• Artificial Intelligence

Background:

• Attention-deficit/hyperactivity disorder (ADHD) is a common neurodevelopmental disorder impacting learning and social interactions.
• Current ADHD diagnosis relies on subjective behavioral assessments, often leading to delays due to clinician shortages and symptom heterogeneity.
• There is a critical need for more efficient and objective methods for early ADHD detection.

Purpose of the Study:

• To investigate if fusing neurophysiological data from multiple tasks can improve ADHD detection accuracy.
• To develop and evaluate a Deep Neural Network (DNN) based fusion model for enhanced ADHD diagnosis.
• To assess the impact of task difficulty and duration on probing brain resources in children with ADHD.

Main Methods:

• Recruited 49 children diagnosed with ADHD and 32 typically developing children.
• Utilized a Deep Neural Network (DNN) based fusion model to integrate neurophysiological data from multiple tasks.
• Collected electroencephalogram (EEG) data during tasks of varying difficulty and duration.

Main Results:

• The proposed multi-task DNN fusion model achieved an 89% separation rate, significantly higher than single data types (81% accuracy).
• The model demonstrated effectiveness in distinguishing between children with and without ADHD.
• Multiple tasks provided essential neurophysiological information for improved ADHD screening.

Conclusions:

• Fusing multi-task neurophysiological data with a DNN model offers a more efficient and accurate method for early ADHD diagnosis.
• This AI-driven approach using multimodal data represents a significant advancement in digital health for ADHD management.
• The findings support the use of advanced computational models for objective clinical screening and intervention.