Highlight
The TriMaster crossover trial rigorously compared four precision medicine strategies to target glucose-lowering therapy in type 2 diabetes, assessing their predictive power for individualized HbA1c response to sitagliptin, canagliflozin, and pioglitazone. Among these, a routine-features treatment-selection model based on clinical parameters significantly outperformed cluster-based and polygenic score approaches in predicting glycemic response and optimizing drug choice. Clinical clusters offered limited benefit confined to one specific subgroup-drug comparison, while genetic-based polygenic scores showed only modest associations. The findings underscore the clinical utility of direct outcome prediction over unsupervised clustering or genetic stratification for short-term HbA1c management.
