Artificial intelligence in psychiatry: A global perspective on research status, trends and clinical applications

Abstract

Artificial intelligence is increasingly used in psychiatric research, yet progress in clinical use and equitable benefit remains uneven. We analyzed 2795 Web of Science publications on artificial intelligence in psychiatry using bibliometrics and BERTopic based semantic topic modelling. The literature has expanded rapidly since 2012, but output and collaboration are concentrated within a limited set of high-income regions and tightly connected author communities. Across the semantic map, methodological development and diagnosis and monitoring applications occupy central positions, while ethics, governance, and implementation research remain sparse and weakly connected to clinical translation. A temporal gap analysis shows that ethics and regulatory research remained consistently scarce, resulting in a persistent structural gap between clinical applications and ethical governance that widened after 2016. Topic hierarchies also indicate limited representation of older adults, digitally marginalized groups, and low resource settings, which may constrain generalizability and widen disparities. These patterns support a shift toward stronger governance, inclusive data practices, and multisite validation, alongside international collaboration and data sharing approaches that prioritize equity.