همایش بین المللی هوش مصنوعی

صفحه اصلی / دومین همایش بین المللی هوش مصنوعی

Artificial Intelligence in ADHD Diagnosis and Treatment: A PRISMA-Aligned Systematic Review

نویسندگان :

Fatemeh Ranjbar¹ Rashin Gholijani Farahani² Mohammad Cheraghi³ Seyed Sepehr Mirhoseyni Nayeri⁴ Niloofar Mirzaei Chahardeh⁵

1- دانشگاه آزاد اسلامی واحد کرج 2- دانشگاه آزاد اسلامی واحد کرج 3- دانشگاه آزاد اسلامی واحد تهران مرکزی 4- دانشگاه آزاد اسلامی واحد کرج 5- دانشگاه آزاد واحد علوم و تحقیقات

کلمات کلیدی :

ADHD،artificial intelligence،machine learning،diagnosis،digital health

چکیده :

ADHD is one of the most prevalent neurodevelopmental disorders worldwide among both children and adults. However, conventional approaches to diagnosis are still plagued by subjectivity, inter-rater variability, and cultural or clinical biases. More recently, Artificial Intelligence has been an attractive avenue through which to develop more objective and data-driven methods for diagnosis. This systematic Review, conducted in accordance with PRISMA guidelines, synthesizes findings from 64 empirical studies published between 2020 and 2025 on a wide variety of AI-enabled approaches, including functional neuroimaging (fMRI), electrophysiological signals (EEG), behavioral sensing technologies, speech analysis, and digital therapeutic tools such as Virtual Reality. Overall, machine learning and deep learning models performed remarkably well in controlled research settings, where internal validation accuracy for fMRI- and EEG-based approaches was often reported between 94 and 99%. However, further detailed review shows that these high values of accuracy are almost exclusively based on internal or single-site validation. These raise highly worrying concerns about overfitting and generalizability in real life. In the therapeutic domain, virtual reality and robotics-based interventions showed high levels of engagement and feasibility, while consistent evidence on the transfer of such benefits to everyday functioning remains scarce. This Review highlights a huge gap between experimental AI models and clinically deployable systems. Cross-site datasets, strict external validation, transparent reporting standards, and robust ethical frameworks for ensuring fairness, safety, and privacy in real-world assessment and treatment of ADHD will be needed to bridge this gap.