A study presented at the APSARD 2026 Annual Conference addressed the validity of telehealth in ADHD diagnosis, specifically an online asynchronous assessment for adults. Barry K. Herman, MD, MMM, chief medical officer of Mentavi Health, presented real-world data on this topic.
Background on Telehealth for ADHD
Telehealth use for mental health care increased significantly during the COVID-19 pandemic, with 46% of adults with ADHD utilizing it. Despite its widespread use, few studies had directly compared telehealth-based diagnostic tools with traditional clinical interviews.
Study Design and Key Findings
The study, a real-world evaluation, compared Mentavi Health's online diagnostic assessment with traditional clinical interviews performed by licensed mental health professionals. Collaborators included Stephen V. Faraone, PhD, Andrew J. Cutler, MD, and Jeffrey H. Newcorn, MD. Key findings indicated a high degree of agreement between the online evaluation and clinical interviews.
Key metrics reported:
- Sensitivity: 80%
- Positive Predictive Value: 95%
- False Positive Rate (Online Assessment): 12%
- False Positive Rate (Clinical Interview): 56%
These results were published in the Journal of Clinical Psychiatry in September 2025. Dr. Herman noted the variability in current ADHD diagnosis approaches, highlighting that many healthcare professionals performing diagnoses may lack sufficient background, training, or expertise. He also emphasized the current absence of widely accessible biomarkers for ADHD, meaning diagnosis relies on clinician judgment against DSM-5 criteria.
Addressing Diagnostic Complexity and AI
The diagnostic tool accounts for the high rate of comorbidity in ADHD patients, as 80% have at least one co-occurring psychiatric condition. It identifies commonly associated mental health conditions, providing data for clinicians' further exploration.
Regarding the role of artificial intelligence, Mentavi Health maintains an AI ethics committee. The company states that all aspects of diagnosis and treatment involve human professionals, ensuring that AI or chatbots do not make diagnoses. The data collected by the asynchronous assessment is used by human clinicians for their clinical judgment.