Applied Machine Learning Method to Predict Children With ADHD Using Prefrontal Cortex Activity: A Multicenter Study...

Area of Science:

• Neuroscience
• Computational Psychiatry
• Biomarker Discovery

Background:

• Attention Deficit Hyperactivity Disorder (ADHD) diagnosis relies on subjective assessments.
• Objective biomarkers are needed to improve diagnostic accuracy and efficiency for ADHD.

Purpose of the Study:

• To develop and validate objective biomarkers for ADHD using machine learning.
• To predict ADHD severity utilizing brain function data and behavioral indicators.

Main Methods:

• A multicenter dataset of 108 children with ADHD and 108 typically developing (TD) children was used for training.
• Support Vector Machine (SVM) machine learning model applied to predict ADHD severity.
• Near-infrared spectroscopy (NIRS) measured prefrontal cortex oxygenation during the reverse Stroop task (RST).

Main Results:

• The SVM model achieved 88.71% sensitivity and 83.78% specificity.
• An overall discrimination rate of 86.25% was obtained in predicting ADHD.
• Verification data from 99 children (62 ADHD, 37 TD) supported the findings.

Conclusions:

• A SVM model utilizing objective data from the RST shows promise as an auxiliary biomarker for ADHD diagnosis.
• This machine learning approach offers a potential objective method to support ADHD diagnosis in children.
• Further research can explore the integration of these biomarkers into clinical practice for ADHD evaluation.