Disease Symptoms Analysis Using Data Mining Techniques to Predict Diabetes Risk

Abstract

Data mining field concentrates on theories, concepts, methodologies and mainly on extraction of useful knowledge from large amounts of data for decision making. During their day to day activities healthcare industry generates large amounts of healthcare information that has not been efficiently used to extract unknown information. Therefore the discovery of interesting and useful information usually remains hidden. Diabetes is a healthcare problem and is increasing at a greater rate with every passing year. If not recognized early, can lead to severe health problems, even in organ failures. Several data mining techniques like clustering, classification, association rule mining are used to identify early symptoms of the diseases and stopping them getting to a chronic level. In this paper, an efficient approach has been designed for prediction of risk of getting diabetes using diabetes database. The approach in this paper used more than one data mining techniques showing enhanced result in disease prediction. The data for diabetes is collected and processed to facilitate the mining process. Firstly, the preprocessed database is mined to extract frequent patterns related to diabetes using FP-Growth algorithm. After that ID3 algorithm approach has been used as the training algorithm to depict the risk of diabetes using a Decision Tree.