Prospective Evaluation of iDECIDE: A Smartphone App for Insulin Bolus Dosing

The investigators will seek approval from Mayo Clinic and Arizona State University Institutional Review Board to recruit 30 type 1 diabetes adult outpatients using insulin pumps and continuous glucose monitoring system from the Endocrinology Clinic at Mayo Clinic Arizona. First, participants will complete a questionnaire to self-assess their confidence to count carbs from food/drinks, personal perceptions on how exercise/alcohol affect their blood glucose, and how they learned about it (e.g., trial and error, education from providers), and compensation techniques used to adjust for exercise performed/alcohol consumed (for instance, adjust for basal rate or adjust insulin boluses). Stored glucose data will be analyzed for the previous 8 weeks prior to entry to establish patient-specific upper and lower bounds for determining excessive blood glucose highs and lows. Second, participants will be randomly assigned to 4 weeks of using their pump bolus wizards to decide insulin boluses and then 4 weeks of using iDECIDE to receive recommendations for insulin dosing, or vice versa. Throughout the 8 weeks, study participants will be asked to wear a wristband heart rate accelerometer that will provide data on time and duration of exercise performed. Also, during the 8 weeks, study participants will use their glucose sensor and pump to measure blood glucose and to deliver basal insulin. During the 4 weeks when participants receive recommendations from iDECIDE, the pump's bolus calculator will be turned off so participants will receive no insulin bolus recommendations from pump. While receiving recommendations from iDECIDE, participants will use the pump to manually deliver insulin boluses and can either use the advice from iDECIDE or override it.

The primary outcomes to compare iDECIDE's insulin dosing algorithm against the pump calculators will be: mean postprandial glucose level and number of events with excessive blood glucose highs and lows. Potential secondary outcomes: mean percentage of time with low glucose level and time spent in the target glucose range.