Machine Learning Stop Signal Test (ML-SST): ML-based Mouse Tracking Enhances Adult ADHD Diagnosis

Abstract

Attention-deficit/Hyperactivity disorder (ADHD) affects the quality of life worldwide. It is commonly diagnosed and studied with specialized questionnaires and behavioral tests. However, in cases of late-onset or mild forms of ADHD, behavioral measures often fail to gauge the deficiencies well-highlighted by questionnaires. This lack of sensitivity in behavioral tests is problematic because it prevents researchers from studying pathophysiology of ADHD ranging from normal to abnormal. To improve the sensitivity of behavioral tests, in the present study we propose a novel version of the Stop-signal task (SST) - a common behavioral test of ADHD - which integrates machine learning and mouse cursor tracking (ML-SST). In one experiment, we compared ML-SST and a standard version of SST (s-SST) in their ability to detect ADHD symptoms in an adult sample. Our results indicate that introducing mouse cursor tracking and ridge regression produces the strongest and most stable associations between questionnaire data and behavioral measures.