Automating self-regulation of insulin-dependent diabetics by using fuzzy logic

The consequences of diabetes mellitus, a metabolic disease, put the society to great expenses. A diabetic himself, especially with respect to the type-I variant (insulin dependent diabetes mellitus), is for the main part responsible for the seriousness of these consequences. With adequate regulation it is even possible to reduce the likelihood of future complications to that of a non-diabetic. The research described in this thesis aims at two different approaches of regulating diabetes by predicting and advising control actions to be employed by a type-I diabetic. The first, modelling the patient specific diabetes mellitus system and deriving a control action is shown to be much too premature to be used sensible and safe. Next, modelling the decision behaviour of a skilled diabetic by means of Fuzzy Logic theory is proposed and implemented in a specific case. Such systems can be used to enlighten discussion between doctor and patient on the one hand, and to support diabetics in determining adequate control actions on the other hand.