LAUSR

The traditional diagnosis of Attention Deficits/Hyperactivity Disorder (ADHD) is through parent-child interviews and observations; therefore, innovative ADHD diagnostic tools that represent this digital era are needed. Virtual reality (VR) is a significant technology that can present a virtual immersive environment; it can provide an illusion of participation in an artificial milieu for children with ADHD. This study aimed to develop an ADHD-VR diagnostic tool construct (Research Domain Construct/RDC) based on the DSM5 ADHD diagnostic criteria, and using the RDC to develop a diagnostic tool with a machine learning (ML) application that can produce an intelligent model to receive some complex and multifaceted clinical data (ADHD clinical symptoms). We aimed to expand a model algorithm from the data, and finally make predictions by providing new data (output data) that have more accurate diagnostic value. This was an exploratory qualitative study and consisted of two stages. The first stage of the study applied the Delphi technique, and the goal was to translate ADHD symptoms based on DSM 5 diagnostic criteria into concrete behavior that can be observed among children in a classroom setting. This stage aimed to gather information, perceptions, consensus, and confirmation from experts. In this study, three rounds of Delphi were conducted. The second stage was to finalize the RDC of the ADHD-VR diagnostic tool with ML, based on the first-stage results. The results were transformed into concrete activities that could be applied in the programming of the ADHD-VR diagnostic tool, followed by starting to input data that were required to build the diagnostic tool. The second stage consisted of more than ten focus-group discussions (FGDs) before it could be transformed into the ADHD-VR diagnostic tool with the ML prototype. First-stage data analysis was performed using Microsoft Excel for Mac. Qualitative data were analyzed using conceptual content analysis with a manifest/latent analysis approach. From the first stage of the study, there were 13 examples of student behaviors that received more than 75% totally agreed or agreed from the experts. The RDC of the ADHD-VR diagnostic tool with machine learning application consisted of three domains and was divided into six sub-domains: reward-related processing, emotional lability, inhibitory, sustained attention, specific timing of playing in order, and arousal. In conclusion, the results of this study can be used as a reference for future studies in a similar context and content, that is, the ADHD-VR diagnostic tool with machine learning based on the constructed RDC.