A novel adaptive basal therapy based on the value and rate of change of blood glucose

Abstract

Background: Modern insulin pump therapy for type 1 diabetes mellitus offers the freedom to program several basal profiles that may accommodate diurnal ariability in insulin sensitivity and activity level. However, these basal profiles do not change even if a pending hypoglycemic or hyperglycemic event is foreseen. New insulin pumps could receive a direct feed of glucose values from a continuous glucose monitoring (CGM) system and could enable dynamic basal adaptation to improve glycemic control.

Method: The proposed method is a two-step procedure. After the design of an initial basal profile, an adaptation of the basal rate is suggested as a gain multiplier based on the current CGM glucose value and its rate of change (ROC). Taking the glucose value and its ROC as axes, a two-dimensional plane is divided into a nine-zone mosaic, where each zone is given a predefined basal multiplier; for example, a basal multiplier of zero indicates a recommendation to shut off the pump.

Results: The proposed therapy was evaluated on 20 in silico subjects (ten adults and ten adolescents) in the Food and Drug Administration-approved UVa/Padova simulator. Compared with conventional basal therapy, the proposed basal adjustment improved the percentage of glucose levels that stayed in the range of 60-180 mg/dl for all 20 subjects. In addition, the adaptive basal therapy reduced the average blood glucose index values.

Conclusions: The proposed therapy provides the flexibility to account for insulin sensitivity variations that may result from stress and/or physical activities. Because of its simplicity, the proposed method could be embedded in a chip in a future artificial pancreatic beta cell or used in a "smart" insulin pump.

Figure 1.

Schematic illustration of basal–bolus therapy. The green line denotes the basal insulin rate during the day, and the three red bars denote boluses for breakfast, lunch, and dinner, respectively.

Figure 2.

Basal gain mosaic. The two-dimensional plane is divided into nine zones, A–I. The different colors denote five predefined gains for the basal rate: 0, 0.5, 1, 1.5, and 2.

Figure 3.

Glucose prediction results. The solid curve is the BG concentration. The dashed curve is the CGM reading. The circle denotes the prediction (every 30 min) of glucose based on CGM for A adolescent 4 and B adult 1. Both subjects consumed fived meals, of which the size and timing are 45, 70, 5, 80, and 5 g and 7:00 AM, 12:00 PM, 4:00 PM, 6:00 PM, and 11:00 PM, respectively.

Figure 4.

Control results for adult 2, where a comparison of glucose response under two therapies is presented in A and a comparison of basal insulin is in B. As can be seen, adaptive basal therapy successfully prevented severe hypoglycemia. Figure 5. Control results for adolescent 9, where a comparison of glucose response under two therapies is presented in A and a comparison of basal insulin is in B. As can be seen, adaptive basal therapy successfully prevented most hyperglycemia.

Figure 5.

Control results for adolescent 9, where a comparison of glucose response under two therapies is presented in A and a comparison of basal insulin is in B. As can be seen, adaptive basal therapy successfully prevented most hyperglycemia.

Figure 6.

Control-variability grid analysis results for 20 subjects: A adaptive basal therapy, with 10% in zone A, 85% in zone B, and 5% in zone C and B fixed basal therapy, with 5% in zone A, 85% in zone B, and 10% in zone C.

Figure 7.

Block diagram of adaptive basal therapy and advisory mode control: A adaptive basal therapy and B advisory mode control. The components in the dashed frame are not used in the advisory mode control.

Figure 8.

Advisory mode evaluation results for subject 1: A CGM values and hyper- and hypoglycemic thresholds and B original basal insulin and suggested basal insulin. The total original basal insulin is 39.4 U, and the total suggested basal insulin is 34.6 U. The following meals and matching boluses were given: at 10:12 am, a meal with 61 g carbohydrates and 10.9 U of bolus insulin were consumed; at 1:02 pm, a meal with 60 g carbohydrates was consumed without bolus; at 3:42 pm, 10 U of bolus was injected; and at 10:21 pm, a meal with 60 g carbohydrates and 6.8 U of bolus were consumed.

Figure 9.

Advisory mode evaluation results for subject 2: A CGM values and hyperglycemic and hypoglycemic thresholds and B original and suggested insulin delivery rates. The total original basal insulin is 13.2 U, and the total suggested basal insulin is 14.8 U. At 8:04 AM, a breakfast with 52.5 g carbohydrates was consumed and no bolus was injected.