A novel hybrid intelligent system with missing value imputation for diabetes diagnosis

Abstract

Recently, diabetes becomes the widespread and major disease in the world. In this paper, we propose a novel hybrid classifier for diabetic diseases. The proposed hybrid classifier named Logistic Adaptive Network-based Fuzzy Inference System (LANFIS) is a combination of Logistic regression and Adaptive Network-based Fuzzy Inference System. Our proposed intelligent system does not use classifiers to continuous output, does not delete samples with missing values, and does not use insignificant attributes which reduces number of tests required during data acquisition. The diagnosis performance of the LANFIS intelligent system is calculated using sensitivity, specificity, accuracy and confusion matrix. Our findings show that the classification accuracy of LANFIS intelligent system is about 88.05%. Indeed, 3–5% increase in accuracy is obtained by the proposed intelligent system and it is better than fuzzy classifiers in the available literature by deleting all samples to missing values and applying traditional classifiers to different sets of features.