Framework for pattern extraction and classification of clinical data

Abstract

A major challenge in healthcare organizations is the extraction of newlineknowledge from the clinical data, which can support the clinician in decision newlinemaking process. Classification is a widely used supervised data mining newlinetechnique for knowledge extraction. The integration of discovered knowledge newlinewith clinical decision making system reduces medical errors, enhances the newlinediagnostic process, decrease practice variation and improves patient s newlinesatisfaction. The performance of a clinical decision making system is mainly newlinebased on characteristics of data and algorithms used for classification. newlineGenerally, clinical data that holds the results of health care examination are newlinedescribed using continuous-valued attributes, which challenges the process of newlinemining in clinical data. Hence, there is a need to pre-process the clinical data newlinebefore mining. This research work aims in pre-processing the continuous newlinevalued clinical data using discretization methods for building a classifier. newlineFurthermore, to enhance the human reasoning in clinical decision making newlineability of the classifier model, fuzzy set theory has been applied in designing newlinethe classifier. The classification model has been evaluated using four clinical newlinedatasets namely, Pima Indians Diabetes (PID) dataset, BUPA Liver Disorder newline(BLD) dataset, Cleveland Heart Disease (CHD) dataset, Chronic Kidney newlineDisease (CKD) dataset taken from UCI machine learning repository. newline newline