Applications of Machine Learning in Traditional Chinese Medicine Quality Control: Methods, Progress, and Challenges

Quality control is central to the safety and consistency of traditional Chinese medicine (TCM), yet the chemical complexity of herbal materials complicates the interpretation of single-marker tests. Machine learning can extract patterns from high-dimensional analytical data and support defined quality-control tasks. This narrative review examines dimensionality reduction, spectral preprocessing, classical and deep models, and knowledge-driven fusion across four application scenarios: authentication and origin traceability, metabolomics-based assessment, processing and manufacturing control, and consistency evaluation of decoction pieces and preparations. Two machine-learning case studies and one process-analytical monitoring example are assessed against a common validation checklist, and the reporting of validation in cited performance studies is summarized. Current limitations include small datasets, distribution shift, and limited independent-domain validation. We outline priorities for process monitoring and for linking analytical predictions to pharmacological and clinical evidence. The review provides a task- and evidence-based guide to applying machine learning in TCM quality control.