Medical Distress Prediction Based on Classification Rule Discovery Using Ant-Miner Algorithm

Enormous data mining techniques were used for disease prediction among which only a few have employed feature selection. The prediction knowledge for disease diagnosis highly depends on the subjective knowledge of the experts. Developing a disease prediction model in time can help us to overcome the medical distress. In this paper, three feature selection strategies namely, HS, MS and TS are devised to obtain the valuable subset of relevant features for reducing the dimensionality of multiple attributes. This work proposed a modified ant-miner algorithm to extract the classification rules from the data. Three bench marked datasets (Cleveland, Pima and Wisconsin) from the UCI machine learning repository were used to analyze effectiveness of the proposed model. The obtained results clearly shows that the modified ant-miner outperforms the other top data mining classification algorithms like the CN2, RBF, Adaboost and Bagging in terms of accuracy. Thus the proposed model is capable of producing good results with fewer features and serves as a suitable tool for eliciting and representing the expert's decision rules with an effective support for solving disease prediction problem.