Gayashani WIA, Herath NDM, Hapuarachchi AS, Jayamaha AR. Bibliometric mapping of the multidimensional landscape of artificial intelligence-based assistance for combating substance use disorders (AI-ACSUDs). Discover Artificial Intelligence. 2026;doi:10.1007/s44163-026-01791-2
This mixed-methods bibliometric analysis examined global trends, methodological maturity, translational gaps, and thematic priorities in artificial intelligence applications for alcohol, cannabis, and substance use disorders (AI-ACSUDs). 4,657 Scopus-indexed publications from 2020 to 2025 were examined, and 2,328 highly cited articles were thematically evaluated. After applying a minimum occurrence threshold of 200, 96 keywords met the inclusion criteria. The keyword analysis identified 4,507 links and a total link strength (TLS) of 404,445. The most influential terms included “human/humans” (6,982 occurrences, TLS = 90,666), “male” (2,591, TLS = 38,569), “machine learning” (816, TLS = 8,885), and “drug dependence” (515, TLS = 6,553). Four major clusters emerged: clinical epidemiology and risk; machine learning and behavioral health; opioid treatment and drug safety; and analytical detection and study design. The analysis revealed that the primary focus of published research changed over time. From 2022, publications shifted toward non-opioids such as cannabis and marijuana. In 2023, synthetic opioids became prominent, with fentanyl appearing 11 times, while “opioid use disorder” appeared 31 times and “substance use” 33 times. By 2024, depression (16 times) and mental health (32 times) reflected stronger connections between substance use and mental health. AI research also progressed from basic machine learning toward deep learning (DL) and natural language processing (NLP). By 2025, scalable AI models, patient safety, and treatment optimization gained attention. Despite this shift, chatbots, generative AI, conversational agents, ChatGPT, and digital phenotyping remained emerging topics. Publications and collaborations were concentrated mainly in North America and Europe. Of importance, two areas that remained critically under-published were ethical and governance issues. Overall, the field is expanding toward digital therapeutics and AI-supported interventions, but methodological convergence, translation, and ethical integration remain important gaps.