A Centroid-based Clustering Approach to Analyze Examinations for Diabetic Patients

Abstract

Health care sector is exploding with data from various streams such as patient history, insurance details, examination histories, drug prescription and many more. This data has immense potential to serve the health care sector in various ways. But due to the huge outbreak, the sector demands powerful and innovative mining tools for extracting useful information from these data. In this paper we propose a novel and explorative approach for data mining the similarities of examinations and medical conditions prevailing among diabetic patients which are in different age groups. This paper, we use a centroid-based clustering algorithm for grouping the data obtained and a decision tree classifier for classification. The decision tree classifier is chosen because by looking at the tree structure all persons who uses the model can easily understand what the model means. For studying different data attributes, the clustering algorithm has been taken in multi-level fashion.