A new clinical decision support system with instance based ensemble classifiers [Örnek tabanli siniflandirici topluluklariyla yeni bir klinik karar destek sistemi]

The main reason of obesity occurring in the future years is strongly related with the lifestyle and eating habits in childhood. this study focuses on developing an urgent precaution system which calculates the obesity risks. an original clinical decision support system (cdss) has been developed by using ensemble classification methods in machine learnings. a questionnaire has been prepared and applied to the patients in the hospitals and to the elementary school students with official permissions in order to construct an original and reliable dataset. extended versions of k nearest neighbors methods are used in voting,bagging,boosting and random subspaces algorithms as base learners. during the experimental studies in the applications of cross validation procedures,successful results have been computed and random subspaces has been chosen as the most successful algorithm with 0.839 mcc (matthews correlation coefficient) scores. with the help of the suggested model to a worldly wide health problem,the future probability of obesity risk for a child might be easily determined. additionally,it has been enabled that some precautions can be taken by responsible people if there is a computed high risk for this child.