BackgroundModern medicine enables dose optimization based on pharmacokinetics-pharmacodynamics (PK-PD) models. In contrast, dose adjustment for traditional Chinese medicine (TCM) formulas lacks quantitative standards and relies largely on empirical practice. This is especially problematic in managing chronic diseases like type 2 diabetes, which require long-term dose titration. Key research bottlenecks include unclear dose-response relationships, a lack of sensitive biomarkers, and the absence of quantitative guidance for dynamic adjustment, hampering scientific decisions on 'when and how much to adjust.' Consequently, developing a scientific methodology for TCM formula dose adjustment is a crucial unmet clinical need.MethodsWe conducted a two-phase study. Phase I (retrospective analysis) pooled patient data from seven multicenter clinical studies. Using multivariate and logistic regression alongside ROC analysis, we screened baseline and on-treatment variables to identify those most predictive of HbA1c reduction. This analysis established the optimal timing (week 4) and threshold (FPG reduction <0.5 mmol/L) for dose adjustment and modeled the magnitude of adjustment required. Phase II was a prospective, randomized, controlled trial that compared the efficacy and safety of a fixed-dose regimen versus a dynamic, response-guided titration strategy (the 'symptom-driven dose adjustment' protocol) based on the rules derived from Phase I.ResultsThe key findings of this study can be summarized as follows: a) A reduction in fasting plasma glucose (FPG) of <0.5 mmol/L at 4 weeks of treatment was established as the critical threshold for triggering a dose adjustment. b) Different TCM formulas (e.g., Gegen Qinlian Decoction (GQD) and Tang-min-ling pills (TM81) exhibited distinct dose-response relationships, indicating that the magnitude of dose adjustment should be determined on an individualized, formula-specific basis. c) Analysis of the dose-ranging clinical data for GQD indicated that a dose increase of approximately 50% is appropriate when the aforementioned indicator threshold is not met. d) The prospective validation study demonstrated that the symptom-driven dose adjustment group achieved superior outcomes in HbA1c, FPG, and other glycemic parameters compared to the fixed-dose group (P < 0.05).ConclusionThe 'symptom-driven dose adjustment' strategy provides a scientific basis for dosing traditional Chinese medicine formulas, enhances clinical efficacy, and facilitates the optimization of treatment regimens.