Abstract / Summary
Background: Attention-deficit/hyperactivity disorder (ADHD) affects over 366 million adults and 139 million children worldwide, yet diagnosis remains fundamentally subjective, relying on clinical interviews, behavioral observations, and rating scales that yield inconsistent results across practitioners and settings. Artificial intelligence (AI), machine learning (ML), and deep learning (DL) offer a paradigm shift toward objective, data-driven diagnosis by detecting complex patterns across neuroimaging, electrophysiology, and digital biomarkers that elude conventional assessment. Although AI-based ADHD research has grown exponentially, no comprehensive synthesis examines the full spectrum of data modalities, validation practices, and clinical translation readiness. This gap limits our understanding of which approaches are most promising for real-world implementation.
Objective: This scoping review will visually map the current evidence on AI-based ADHD classification with respect to predictive accuracy, data forms, data features, generalizability, and interpretability of models.
Methods: This scoping review will use the Joanna Briggs Institute approach to scoping reviews and follow the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines. Five electronic databases (IEEE Xplore, Scopus, PubMed, Web of Science, and ACM Digital Library) will be systematically searched for peer-reviewed studies published between January 2019 and April 2026. Empirical studies in English involving the use of AI, ML, DL, or explainable AI to diagnose ADHD with a total sample size greater than 100 participants (ADHD and control groups combined) and a non-ADHD control group (individuals with typical development or healthy individuals) will be included. Two independent reviewers will screen the titles, abstracts, and full texts, and any conflicts will be resolved through either discussion or arbitration. A standardized Microsoft Excel template will be used to extract data that will include the following: bibliographic data, data modalities, data models, data validation approaches, performance metrics, and explainability approaches. Thematic and narrative analysis will be used to synthesize findings on 4 research questions that will address model performance, data modality contributions, data characteristics, and interpretability methods.
Results: The scoping review began in December 2025. Analysis and screening are in progress, with the scoping review expected to be completed and submitted for publication in June 2026.
Conclusions: This scoping review will provide the first comprehensive synthesis of AI-, ML-, and DL-based ADHD diagnostic classification studies, mapping the evidence across data modalities, validation practices, interpretability methods, and clinical translation readiness. Findings will inform future methodological standards and support the translation of AI-based diagnostic tools into clinical practice.