A groundbreaking study published in Translational Psychiatry has made significant strides in developing an artificial intelligence (AI) model that can accurately diagnose attention-deficit/hyperactivity disorder (ADHD) in adults using virtual reality, eye movement data, and self-reported symptoms. The innovative approach combines multiple assessment modalities to improve diagnostic accuracy, offering a promising solution for a condition often misdiagnosed or undiagnosed.
The Challenges of Adult ADHD Diagnosis
Adults with ADHD face unique challenges when it comes to diagnosis. Unlike childhood diagnoses, which rely on clinical interviews and retrospective self-reports, adult diagnoses require more objective measures. Unfortunately, no established biomarkers or lab tests confirm an ADHD diagnosis, leading to misdiagnosis and delayed treatment.
A Multimodal Approach to Diagnosis
Researchers aimed to develop a more accurate diagnostic tool by combining performance on a sustained attention task with eye tracking, head motion measurements, electroencephalography (EEG), and real-time self-reports. Participants completed the task in a simulated seminar room using virtual reality, where distractions mimicked everyday interruptions.
"ADHD is a complex and heterogeneous disorder, and no cognitive tests or biomarkers exist that can accurately detect it," said co-first author Benjamin Selaskowski. "Combining multiple assessment modalities may improve diagnostic accuracy."
The study consisted of two phases: training data collection from 50 adults (25 with ADHD and 25 without) and predictive accuracy testing on a separate group of 36 participants (18 with ADHD and 18 without).
AI-Powered Diagnosis
The machine learning model was trained to identify patterns across different types of data that were most predictive of ADHD. The researchers used a statistical method called maximal relevance and minimal redundancy (MRMR) to select variables strongly related to the diagnosis and relatively uncorrelated with each other.
Out of 76 total features, the optimal model used only 11 to achieve high performance. These features came from four data sources: self-reported symptoms, eye tracking, task performance, and head movement. Gaze wandering, reaction time variability, and head motion during the task were among the most important predictors.
When applied to the independent test set, the model achieved an overall accuracy of 81%, with a sensitivity of 78% and specificity of 83%. This means it correctly identified 78% of ADHD cases and 83% of non-ADHD cases.
The Power of Virtual Reality
The use of virtual reality was crucial in this study. Traditional attention tests often take place in quiet, sterile environments that do not reflect the noisy, distracting situations people with ADHD face. By placing participants in a more realistic environment and introducing distractions, the researchers captured behaviors that may not emerge in standard tests.
"This study shows that combining multiple types of information can effectively help identify adults with ADHD," explained co-first author Annika Wiebe. "These findings highlight the potential of using a multi-method assessment to improve the accuracy and objectivity of ADHD diagnosis in adults."