Hybrid Multistage Fuzzy Clustering System for Medical Data Classification

Abstract

Due to the rapid development in technology nowadays, massive amount of data are available. In medicine, decision making is entirely based on the hidden information in these massive data. For that reason, data mining and machine learning technologies provide powerful tools for knowledge discovery within data. Two main techniques are used interchangeably: clustering and classification. In machine learning, clustering is an unsupervised learning technique while classification is a supervised learning method. These techniques are capable of extracting useful patterns and information which aid the process of data analysis and clinical decisions. This research presents a recent study of these techniques in the medical field during the past five years. Moreover, this paper proposes a hybrid multistage fuzzy clustering system applied to medical data classification. In the proposed system, two fuzzy clustering algorithms specifically FCM and GK were initially employed to obtain the membership values. These weights are then used in the second stage of the system as additional informative features to improve the classification process completed by SVM algorithm. Wisconsin Breast Cancer dataset, real-world application, obtained from UCI were used in the experiments. The results of the experiments show that the additional weights further improve the classification accuracy with 99.06% and 100% sensitivity.