Digital Twin Metabolic Flux Profiling Enables Precise Prediction of Therapeutic Outcomes in Type 2 Diabetes

Abstract Treatment response in Type 2 Diabetes Mellitus (T2DM) is highly heterogeneous and difficult to predict using conventional clinical markers. Here, we apply digital-twin modeling of generalised metabolic fluxes (GMFs) to 60 T2DM patients treated with Dapagliflozin, Metformin, or their combination. GMF profiling revealed distinct post-treatment metabolic signatures, with Dapagliflozin exerting stronger effects on hepatic and renal fluxes, whereas Metformin and combination therapy strongly modulates haemoglobin glycation. Within each treatment arm, GMF-based clustering stratifies responders and non-responders, with baseline creatinine concentration emerging as a key determinant of glycemic benefit. Proteomic analyses corroborated these sub-groups, showing concordant differences in pathways linked to insulin resistance and inflammation. Longitudinal GMFs further captured the waning glycemic durability of Metformin through progressive increases in Glucose→HbA1c flux. Together, these results establish GMF digital twins as a mechanistic framework to dissect drug-specific effects, stratify heterogeneous responses, and project therapeutic durability in T2DM, offering a new avenue for precision metabolic medicine.