Individuals with Type 1 Diabetes (T1D) require exogenous insulin to keep their blood glucose concentration in a safe euglycemic range, because of the absent internal insulin secretion caused by the autoimmune destruction of pancreatic beta-cells. As a consequence, the quality of glycemic control in T1D is heavily dependent on multiple daily treatment decisions by the patients, which are complicated by a wide variety of factors influencing insulin demand (e.g., circadian rhythms, physical activity, food, stress, etc.). Insulin sensitivity (SI) is a key metabolic parameter in diabetes as it informs on how sensitive the body is to the effects of insulin. In general, if someone has higher SI, the amount of insulin required to lower his blood glucose levels is smaller than that needed by someone who has low sensitivity. However, SI levels within the same person are not constant, and fluctuations of SI happen very frequently in the life of subjects with diabetes, making insulin dosing very difficult to tune.
In this context, the aim of this research project is to develop an SI-informed insulin bolus calculator, with the aim of tailoring the insulin dose to the individual's insulin need at the time the bolus is administered. The SI-informed bolus calculator relies on a Kalman filter-based algorithm which uses continuous glucose monitoring (CGM) data, insulin, and meal records to estimate SI. For each subject, a 24-hour SI profile is computed using data collected over several days of monitoring, and the optimal bolus is then computed by adjusting the standard insulin dose by the ratio between usual SI (from the profile) and real-time SI of the individual at the time the bolus is administered. In this way, if the real-time SI is larger/smaller than the profile SI at that time of day, the insulin dose will be reduced/incremented accordingly.
The study is thus designed as a single-center randomized clinical trial targeting completion of 15 subjects, who will undergo a 28-day at home Data Collection Period followed by two 24-hour admissions (Control and Experimental Admission) performed in random order in a semi-controlled environment (i.e., hotel). The Data Collection is meant to collect data needed to build the 24-hour SI profile for the subject. During the admissions, subjects will undergo a 45-minute afternoon exercise session designed to alter the late-afternoon/evening SI. The dinner meal will then be controlled, and the postprandial glycemic control obtained using the standard bolus calculator (Control Admission) will be compared to the control obtained in response to the optimized SI-informed bolus calculator (Experimental Admission). Metrics computed on CGM data will be compared between the two admissions, including mean blood glucose, time above 250 and 300 mg/dL, time below 70 and 54 mg/dL, and time in 70-180 mg/dL, the primary outcome being the postprandial exposure to hypoglycemia as measured by the Low Blood Glucose Index (a glycemic variability indicator which summarizes the number and extent of low blood glucose events in one single number). If successful, this study will provide a novel, data-oriented paradigm for insulin dosing in T1D.
