Optimal feedback glycaemia regulation in diabetics.

Abstract

A method is described to set up the parameters of a feedback-controlled artificial endocrinous pancreas (a.e.p.) at the optimal values for the individual patient. The method has been developed to reduce the insulin consumption that, in diabetics feedback-controlled by automatic a.e.p., is often higher than requested in the usual discontinuous therapy. The parameter values are obtained by a digital computer procedure based on feedback optimal control concepts. The procedure includes two steps: (a) parameter estimation of a mathematical model of the patient, subjected to a prolonged glucose load and (b) evaluation of some settable a.e.p. parameters that minimise a square integral performance index in the glycaemia excess and insulin infusion rate. This evaluation is performed on the set of patient model and a.e.p. equations, simulated on a digital computer. Experimental results, obtained from eight patients subjected to a one-day controlled diet and feedback connected to the a.e.p., with coefficients both set up following the manufacturer’s suggestions and computed as described above, show an appreciable reduction in insulin consumption and an improved glycaemia response.