Background The modernization of Traditional Chinese Medicine (TCM) has long faced significant bottlenecks, including an unclear material basis of efficacy, subjective diagnostic methods, and a lack of standardized evaluation. Artificial Intelligence (AI) offers promising tools to address these bottlenecks through data-driven approaches. Methods This scoping review systematically synthesizes literature up to 2025, obtained via searches in PubMed, Web of Science, and CNKI. Studies applying AI/ML techniques to TCM problems were included and analyzed across four key domains. Applications & Results AI demonstrates substantial potential in reshaping TCM: (1) Drug Discovery & Formula Optimization: Machine learning models enable high-throughput compound screening and formula recommendation, though often lack TCM-specific design. (2) Mechanism Elucidation: Knowledge graphs and multi-omics data integration help decode the "multi-component, multi-target" actions of TCM formulas. (3) Diagnostic Innovation: Deep learning and large language models (LLMs) demonstrate promising accuracy in tasks such as tongue image analysis and syndrome differentiation, in some studies approaching expert-level performance, thereby aiding standardization efforts. (4) Quality Control: AI-powered spectral and image analysis facilitates rapid, non-destructive assessment of herbal medicine quality. Challenges Critical barriers remain, including data heterogeneity, poor model interpretability ("black-box" issue), a significant translational gap, and a lack of interdisciplinary talent. Conclusion & Future Directions Although most current applications adapt generic AI algorithms, future progress hinges on developing TCM-aware models (e.g., symbolic-neural hybrids), establishing standardized TCM data ontologies, and conducting rigorous clinical validation. The deep integration of AI and TCM holism paves a new path for data-driven, precise TCM and contributes to a global integrated medical system.
董介正 et al. (Mon,) studied this question.