Assessment of an algorithm for the artificial B-cell using the normal insulin-glucose relationship in diabetic dogs and men.

Insulin secretion rates after glucose loading were calculated from peripheral venous IRI concentrations considering half life and distribution space of exogenous insulin in normal men and dogs. The coefficients of multiple linear regression analysis between insulin secretion rates and plasma glucose (level and order and rate of change) were used as algorithm parameters in glucose-controlled insulin infusions. These were carried out in each dog based on individual estimations before the induction of diabetes but in the diabetic patients based on values derived from a group of normal subjects. Using this formula, nearly normal patterns of glucose and of insulin were observed in diabetic men and dogs under basal conditions and after IV glucose loading but not after meals. This algorithm enables selection of the parameters prospectively. The effect of a parameter combination depends on insulin sensitivity and it should be appropriately adapted. In the diabetic patients there was no predictable influence of the brittle or stable characteristics of the disease nor of insulin antibodies on the glucose curves obtained with glucose controlled insulin infusions.