Data-driven profiles of attention-deficit/hyperactivity disorder using objective and ecological measures of attention, distractibility, and hyperactivity

Abstract

In the past two decades, the traditional nosology of attention-deficit/hyperactivity disorder (ADHD) has been criticized for having insufficient discriminant validity. In line with current trends, in the present study, we combined a data-driven approach with the advantages of virtual reality aiming to identify novel behavioral profiles of ADHD based on ecological and performance-based measures of inattention, impulsivity, and hyperactivity. One hundred and ten Spanish-speaking participants (6-16 years) with ADHD (medication-naïve, n = 57) and typically developing participants (n = 53) completed AULA, a continuous performance test embedded in virtual reality. We performed hybrid hierarchical k-means clustering methods over the whole sample on the normalized t-scores of AULA main indices. A five-cluster structure was the most optimal solution. We did not replicate ADHD subtypes. Instead, we identified two clusters sharing clinical scores on attention indices, susceptibility to distraction, and head motor activity, but with opposing scores on mean reaction time and commission errors; two clusters with good performance; and one cluster with average scores but increased response variability and slow RT. DSM-5 subtypes cut across cluster profiles. Our results suggest that latency of response and response inhibition could serve to distinguish among ADHD subpopulations and guide neuropsychological interventions. Motor activity, in contrast, seems to be a common feature among ADHD subgroups. This study highlights the poor feasibility of categorical systems to parse ADHD heterogeneity and the added value of data-driven approaches and VR-based assessments to obtain an accurate characterization of cognitive functioning in individuals with and without ADHD.

Topics & Concepts