Models & Optimisation and Mathematical Analysis Journal
Volume 8, Numéro 1, Pages 14-21

Application Of Artificial Neural Networks Models In Diabetes Mellitus Classification

Authors : Haddad Fatima Zohra . Benbouras Mohammed Amine .

Abstract

Diabetes mellitus is one of the most worrying chronic diseases. It results from disturbances in blood sugar levels causing many complications and sometimes leads to death. International Diabetes Federation has stated that the number of diabetics is rising and could up to 642 million in 2040. The rapid development of information technology has imposed new advanced methods entitled “learning machine” or “artificial intelligence techniques” which have led to impressive results in the medical field. Based on this background, this study contributes to classify diabetes, by the mean applied of artificial neural networks ‘ANN’ method. The idea is based on the application of ANN on the pima Indian diabetes database on 3 cases, according to the number and the type of selected features "attributes". The results show high accuracy of the model with all attributes (92.3%) and without Diastolic blood pressure (92.6%). The proposed ANN model composed by two hidden layers ensures better predictability in data learning and yields data prediction values better than the ones published in previous studies.

Keywords

Diabetes Mellitus ; Artificial Neural Networks (ANN) ; Pima Indian Diabetes (PID) ; Classification