Predicting Rare Classes of Primary Tumors with Over-Sampling Techniques

Abstract

The discovery of hidden biomedical patterns from large clinical databases can uncover knowledge to support prognosis and diagnosis decision makings. Researchers and healthcare professionals have applied data mining technology to obtain descriptive patterns and predictive models from biomedical and healthcare databases. However, clinical application of data mining algorithms has a severe problem of low predictive accuracy that hamper their wide usage in the clinical environment. We thus focus our study on the improvement of predictive accuracy of the models created from the data mining algorithms. Our main research interest concerns the problem of learning a tree-based classifier model from a multiclass data set with low prevalence rate of some minority classes. We apply random over-sampling and synthetic minority over-sampling (SMOTE) techniques to increase the predictive performance of the learned model. In our study, we consider specific kinds of primary tumors occurring at the frequency rate less than one percent as rare classes. From the experimental results, the SMOTE technique gave a high specificity model, whereas the random over-sampling produced a high sensitivity classifier. The precision performance of a tree-based model obtained from the random over-sampling technique is on average much better than the model learned from the original imbalanced data set.