Moor

Optimal doses of artificial insulin (further – insulin) depend on a variety of factors. Adjustment of insulin dosages is a sophisticated task and can be far too complex for most patients. In order to simplify the solution to the problem the Continuous Glucose Monitoring System (CGMS) and continuous subcutaneous insulin infusion systems (insulin pumps) were developed.Blood glucose level (BG) control systems based on CGMS and insulin pumps are being actively developed. Algorithmically, these systems consist of two subsystems: the BG forecasting subsystem and the optimal dose determination subsystem. In the long run this research aims to synthesize the former of the two subsystemsThe problem of BG prediction of patients with insulin-dependent diabetes is considered in this paper. An approach to solving this problem by using adaptive neuro-fuzzy inference system ANFIS is proposed.  The results of research of efficiency of the method are presented. It was shown that ANFIS provides us with high quality of prediction when used on relatively short periods of time.

A direct adaptive method of multicriteria optimization based on neuro-fuzzy approximation of decision maker’s function is considered in this paper. Some results of the investigation of the efficiency of the method for solving two- and three- criteria test problems are presented.

A study of application of different optimality criteria convolutions in problems of multiobjective optimization. The article shows some results of comparative analysis of the convolutions effectiveness for solving multiobjective optimization test problems.

The paper considers the direct adaptive method for multi-criteria optimization based on fuzzy approximation of decision maker’s utility function. We provide some results of method effectiveness research for solving two- and three-criteria test problems.