Artificial Intelligence in Eating Disorders: A Narrative Review of Detection, Digital Interventions, and Implementation Challenges

Abstract

Eating disorders (EDs) are severe psychiatric conditions with rising prevalence, yet access to specialised care remains limited due to resource-intensive assessments. This narrative review critically evaluates the evidence for AI-driven detection and intervention, discusses the significant barriers to real-world deployment, and outlines future directions for the safe, equitable, and effective integration of these powerful tools into clinical practice for individuals with EDs. A systematic search of PubMed, PsycINFO, Scopus, IEEE Xplore, and Google Scholar (inception–March 2026) identified 43 peer-reviewed English-language articles, including original studies, systematic reviews, and policy reports. Data were thematically synthesised across four domains: (1) AI-based screening/risk prediction, (2) comorbidity detection, (3) digital interventions, and (4) ethical/implementation challenges. AI shows promise in early ED screening through electronic health records, linguistic analysis, and social media data. However, detection of psychiatric comorbidities such as depression, anxiety, and obsessive-compulsive disorder remains variable and requires further validation. Chatbot-assisted interventions and smartphone-based monitoring are emerging as scalable tools for symptom tracking and delivering cognitive-behavioural content, potentially improving care continuity. Major implementation barriers persist, including data privacy concerns, algorithmic transparency issues, and low clinician acceptance due to liability fears, lack of interpretability, and poor workflow integration. AI holds considerable potential to enhance ED care through earlier detection and expanded access to digital interventions. Realising this potential requires rigorous prospective validation, clear ethical guidelines, and collaborative frameworks involving clinicians in AI design and oversight to ensure these tools complement, not replace, clinical judgment.