Knowledge and data engineering: Fuzzy approach and genetic algorithms for clustering β-Thalassemia of knowledge based diagnosis decision support system

Abstract

Fuzzy approach plays an important role in Knowledge and Data Engineering, especially for improving the clustering performance in special problems such as medical diagnosis. This technique is not only a powerful method for clustering special tasks but also a useful technique in various areas. Among hybrid approaches of KDE, in this paper, Fuzzy approach and GAs were selected to cluster several transformed β-Thalassemia variables in which this disease is the common genetic disorder found around the world. According to the genetic counselling problems of this disease in Thailand and other countries, the Knowledge Based Diagnosis Decision Support System for Thalassemia was constructed to reduce these problems. The comparison of clustering results of using Fuzzy approach and hybrid techniques on various β-Thalassemia data sets and expert opinion are presented that K-Means clustering obtains the best result with the RMSE 13.0077 from unrecoded variables, on the other hand Fuzzy C-Mean and Fuzzy-GAs obtain the RMSE 13.6235 and 14.3527 from recoded variables, respectively. As these obtained results, other clustering and classification algorithms will be used to improve the results of KDE techniques for implementing Thalassemia Expert System in the future. © 2013 ISSN 1881-803X.