Extracting Rules for Diagnosis of Diabetes Using Genetic Programming

Background: Diabetes is a global health challenge that cusses highincidence of major social and economic consequences. As such, earlyprevention or identification of those people at risk is crucial forreducing the problems caused by it. The aim of study was to extract therules for diabetes diagnosing using genetic programming.Methods: This study utilized the PIMA dataset of the university ofCalifornia, Irvine. This dataset consists of the information of ۷۶۸ Pimaheritage women, including ۵۰۰ healthy persons and ۲۶۸ persons withdiabetes. Regarding the missing values and outliers in this dataset, theK-nearest neighbor and k-means methods are applied respectively.Moreover, a genetic programming model (GP) was conducted todiagnose diabetes as well as to determine the most important factorsaffecting it. Accuracy, sensitivity and specificity of the proposed modelon the PIMA dataset were obtained as ۷۹.۳۲, ۵۸.۹۶ and ۹۰.۷۴%,respectively.Results: The experimental results of our model on PIMA revealed thatage, PG concentration, BMI, Tri Fold thick and Serum Ins wereeffective in diabetes mellitus and increased risk of diabetes. Inaddition, the good performance of the model coupled with thesimplicity and comprehensiveness of the extracted rules is also shownby the experimental results.Conclusions: GPs can effectively implement the rules for diagnosingdiabetes. Both BMI and PG concentration are also the most importantfactors to increase the risk of suffering from diabetes.

Keywords:

Diabetes , PIMA , Genetic programming , KNNi , K-means , Missing value , Outlier detection , Rule extraction