Bayesian probability approach to ADHD appraisal

Area of Science:

• Neuroscience
• Psychology
• Biostatistics

Background:

• Accurate diagnosis of attention-deficit hyperactivity disorder (ADHD) is critical due to significant psychosocial sequelae associated with undiagnosed cases.
• Traditional ADHD diagnosis relies on subjective symptom history, lacking a definitive single test.
• ADHD is a complex disorder with physiological underpinnings and multifactorial etiology.

Purpose of the Study:

• To introduce a novel stochastic method using a Bayesian interface for quantifying and assessing ADHD.
• To demonstrate the method's ability to integrate diverse psychometric and physiological data into a unified diagnostic instrument.
• To establish a framework for meta-analysis of ADHD studies using disparate data and tests.

Main Methods:

• A Bayesian interface was developed to combine various psychometric tests and physiological markers.
• The method refines an individual's probability of ADHD at each assessment step.
• A meta-analysis framework was created to calibrate disparate study data onto a unified probability scale.

Main Results:

• The Bayesian method achieved 100% correct classification in a small study of college students with ADHD and matched controls.
• Individual assessments in the pilot study were incapable of perfect classification.
• Meta-analysis of five studies combining behavioral rating scales and EEG data (56 ADHD, 55 controls) demonstrated the method's utility.

Conclusions:

• The proposed Bayesian stochastic method offers a more accurate and objective approach to ADHD diagnosis.
• This integrated approach enhances diagnostic precision compared to individual assessment tools.
• The framework facilitates robust meta-analysis, unifying findings from diverse ADHD research.