Exploring the directions of artificial intelligence in good health and well-being (SDG3) using big data and LDA topic modeling

Artificial Intelligence (AI) holds significant potential for advancing Sustainable Development Goal 3 (SDG3)—Good Health and Well-being—yet the field remains fragmented across numerous topics and disciplines. In this study, we apply Latent Dirichlet Allocation (LDA) to a final corpus of 60,010 Scopus abstracts after filtering, extracting k = 160 latent topics (selected via metric-based tuning; see Appendix A) and organizing them into a process-oriented, Health Technology Assessment–inspired framework that links Drivers, AI Infrastructure and Methods, Implementation, and Results. Key findings include dominant research streams in disease diagnostics (e.g., breast cancer, cardiovascular disease), personalized treatment, and automation, alongside the emergence of large language models (LLMs) like ChatGPT. Geographical mapping highlights Asia, North America, and Europe as research hubs, while underexplored areas such as AI in social media and student education are identified. We also introduce a quadrant-based trend analysis to distinguish “niche excellence” from “leading research areas” and chart short-versus medium-term dynamics. This methodological contribution not only offers a comprehensive “scientific map” of AI–SDG3 research but also provides a scalable blueprint for mapping AI's role across other SDGs and guiding future theory-driven and policy-relevant investigations.