Towards ecological validity when assessing ADHD symptoms: Patterns in automatically collected, real-world PC activity data

Neuropsychological tests assessing attention and executive function (EF) in individuals with ADHD demonstrate little to no association with real-world ratings of ADHD behaviors. To address this critical gap, this study developed metrics that can analyze automatically collected data to measure levels of attention, motivation, and effort in emerging adults with ADHD. Specifically, we used virtual reality to simulate a study space and collect in-the-moment computer data activity while university students with ADHD (N = 21; 38 % female) engaged in 12 sessions (total 180 h) of real-world tasks. To identify common sequences we performed a qualitative analysis of this work session data (i.e., descriptive window titles, input levels, and window switches), resulting in four themes representing positive and negative work activity patterns. From these themes we derived four metrics, and a quantitative analyses showed that two predicted behavioral indices of attention, effort, and motivation with effects in the moderate range. To our knowledge, we are the first group to design and test such an approach, as well as validate identified computer metrics to behavioral indices of attention and EF. Given the automated nature of computer data collection and analysis, this approach represents a scalable, novel method for ADHD assessment and treatment. • Production of a dataset consisting of almost 200 h of user activity data from people with ADHD. • Production of themes (thematic analysis) detailing user activity data of students with ADHD. • Production of four quantitative metrics derived from the extracted themes. • Validation of the quantitative metrics with the self-reported evaluations from the study.