DIABETEX decision module 2--calculation of insulin dose proposals and situation recognition by means of classifiers

Current research in the field of medical decision making tries to represent and to analyse complex, uncertain and complicated situations. The first version of DIABETEX, which is a decision support system for the treatment of diabetic out-patients, accepts the challenge to overcome these difficulties. It includes a network of rules on the basis of known glucose-insulin relationships under different situations. The insulin dose for type I diabetic patients is suggested accordingly. In this, the application of special cybernetic methods offers the chance to overcome complexity, uncertainty, fuzzyness and incompleteness of data. Two methods of classification are presented to complete the DIABETEX decision unit: (1) the Bayes' classification is used in the calculation of insulin doses for type I diabetic patients on multiple subcutaneous insulin injections considering the basis-bolus concept; (2) fuzzy classification is employed in separating 'normal diabetic days' from days with information on special situations such as exercise, illness, menstruation on the one side, from stress, hypoglycaemia etc. on the other.