Investigation on prediction of disease in health care using data mining

Abstract

Data mining is the study of growing patterns used to predict the probability of future events. Researchers use solutions for data mining such as multidimensional databases, machine learning, soft computing, visualization of information and statistics. Data mining was used to forecast patient details in each category of disease prediction.The classification accuracy on the disease prediction is most important in the medical field due to the processing of huge critical data. In order to have better classification accuracy, the size of the dataset has to be reduced on employing the feature selection method as it is considered crucial importance of this proposed research work. In order to carry out the research, a new framework has been constructed to classify and predict various forms of diseases from the information collected in the real bench mark dataset. This framework has been composed of unique feature selection models and novelclassifiers. Hence these are considered as methodology of the research. Initially, missing values and imputations carried out by data discrimination method. Noise is filtered using data polishing method.The proposed feature selection algorithm dynamically filters the feature of the dataset and generates the subset of the feature for classification. The proposed feature selection model reduces the feature pool for effective validation and leniency during classification. It is developed to distinguish the spurious candidatesand to predict the health status of patient. Secondly, the classification of Pimadatasetandthe leukemia disease in addition with microarray data is carried out after feature selection process. Pima diabetes is the commonest chronic disease which affects one third of the population.The necessity of the Pimadisease classification is increased for diagnosis that would lead prevalence of other disease. Leukemia is most malignant cancer among the peoples. In order to categorize and surveillance of the patient diseasedata for diagnosing is a major objective in this work.A