A computer-implemented system (10) and method (30) for managing diabetes through cloud computing with circadian profiles (61) is provided. A plurality of records containing a circadian profile (61) is stored in a database (14, 25) maintained in a cloud computing infrastructure (21). The circadian profile includes meal period categories (53), typical measurements of pre-meal and post-meal self-measured blood glucose, and diabetes medication (55) dosed during each of the meal period categories (53). An executable application (15, 28) is used to model predicted blood glucose levels, including collecting the self-measured blood glucose measurements, determining and visualizing expected blood glucose values (83, 88, 89, 111, 112) and their predicted errors (84), propagating a suggested incremental change in dosing of the medication (55), and adjusting the visualized expected blood glucose values (83, 88, 89, 111, 112) and their predicted errors (84) based on pharmacodynamics of the medication (55).
