Why traditional medicine trials are hard
Individualised pattern-based prescriptions resist standard parallel-group RCT design. Sham acupuncture controls activate non-specific effects. Multi-component formulas defy single-mechanism explanation. Students learn the design alternatives — N-of-1 trials, pragmatic cluster designs, adaptive enrichment, and the SPIRIT-TCM and CONSORT-TCM extensions.
Statistics and reproducibility
Frequentist and Bayesian frameworks taught in parallel. Pre-registration via OSF, code-and-data archiving, and reproducibility checks are required for all student projects. R / Python / Stan are the working languages.
Reporting and publication
Authors must understand peer review from both sides. Senior students serve as reviewers for junior projects under faculty oversight, learning publication ethics and reviewer conduct alongside their own writing.