Classification of ADHD and non-ADHD using AR models

摘要

Accurate and confident diagnosis of ADHD is highly dependent on subjective observations. Several quantitative methods have been proposed, posing it as a two-class classification problem (ADHD and non-ADHD). However, the results have not made it past the research stage yet, as misclassification rates must be close to 0%. This study aims to discriminate ADHD and non-ADHD subjects using autoregressive models, with a high level of accuracy (85-95%). In addition, a confidence metric is proposed, expressing with how much confidence the classification of ADHD and non-ADHD subjects is made.