A computer-implemented method is presented for recommending insulin dosage adjustments for a patient having diabetes. The method includes: identifying a plurality of bolus events from patient data； grouping bolus events having a recommended bolus dosage substantially equivalent to the amount of administered insulin into a first subset of bolus events； determining a bolus outcome for each of the bolus events in the first subset of bolus events, where the bolus outcome is expressed in relation to a target range of blood glucose values and is selected from a group including above the target range, in the target range and below the target range； determining whether one of the bolus outcomes is predominant amongst the bolus events in the first subset of bolus events； and generating a recommendation pertaining to insulin dosage for the patient in response to a determination that one of the bolus outcomes is predominant.
