Assessing Risk of Diabetes Mellitus

Diabetes is a non-communicable disease which is affecting the growth of developing countries. Our aim is to prevent diabetes by deducting it in the earlier stage so that people take treatment according to it, this can be done by examine the electronic medical record of a patient to discover set of risk factors by applying association rule mining methods. The Electronic Medical Record is very large which provides many rule set as result when association rule mining is used, so in order to summarize rules we go for Bottom-up-summarization algorithm. Rule set summarization techniques such as RPC, APRX-collection and BUS are applied to compress original rule set commonly available in Electronic Medical Record (EMR) system, then to predict the relative risk of diabetes millets as high risk, medium risk and low risk by using the K-Nearest Neighbor. RPC is a Relative patient coverage which can be extracted from status and follow-up patient record. K-Nearest Neighbor is a non-parametric method and they are mainly used for both classification and regression but here we use it for classification where the input will be training data.