Enhanced K means clustering for patient report outcomes

Abstract

Data Mining has become one of the most exciting and fastest growing fields in Information Technology due to an unprecedented growth-rate at which data is being collected through the World Wide Web and stored electronically, to be applicable in almost all fields of human endeavor. Recently, growing attention has been paid to a clustering technique, as Clustering is a very intensive research topic due to its numerous applications. Online retrieval or querying of useful information from these very large data sets for Medical diagnosis in patient report outcomes of hospital information systems (HIS), Biomedical, DNA analysis etc, is becoming an increasing scientific challenge. As a result there is an urgent need for sophisticated tools and techniques that can handle new fields of data mining like information retrieval for Medical diagnosis in patient report outcomes of hospital information systems (HIS) for analysis. With an increasingly large database being available on the internet from multiple database environments; there is a growing opportunity for a global Knowledge discovery in data sets, to process and extract useful knowledge using fully automated techniques. Thus, it is expected that data mining methods will find interesting patterns from large data sets, because practically human beings cannot deal with such a huge amount of data. In the field of medicine, clustering diseases, cures for diseases, or symptoms of diseases can lead to very useful taxonomies. In this thesis the K-mean algorithm is enhanced by Automatic K-value generation for patient reports in HIS. Since it is a medical report data integrity problem with tuning of outlier deduction, the process is carried out by an enhanced K-Means. In the last part of the thesis, global optimality is the focus. Therefore, the overall goal of this thesis is to cover Automatic K, tuning of outlier deduction and Multi-level Multi-label Clustering on diabetes patient report outcomes (PRO). newline newline newline