A novel automated fuzzy model for diabetes mellitus

Finding an expert fuzzy model for glucose-insulin system seems to be essential because this model is always changeable according to parameters such as body weight, individual age, time and numbers of meals, physical activities, and etc. In this paper we try to obtain a fuzzy model for diabetes mellitus. At first a certain diet is introduced and then the amount of carbohydrate in each meal is calculated, then by introduced diet the amount of blood glucose and insulin as outputs of diabetes mellitus system are determined. Although there are some models about glucose-insulin system but in this paper we use automated method, RLS (Recursive Least Squares) algoritm, in order to find a fuzzy relation between inputs and outputs of system for producing a fuzzy model of glucose-insulin system. At last the performance of obtained model with regard to the same inputs is compared with responses of 21st order metabolic model of Sorensen.