Looking for better combination of biomarker selection and classification algorithm for early screening of ovarian cancer

Abstract

This paper demonstrates and evaluates the classification performance of the optimal biomarker combinations that can diagnose ovarian cancer under Luminex exposed environment. The optimal combinations were determined by T Test, Genetic Algorithm, and Random Forest. Each selected combinations' sensitivity, specificity, and accuracy were compared by Linear Discriminant Analysis (LDA) and k-Nearest Neighbor (k-NN). The 8 biomarker data used in this experiment was obtained through Luminex-PRA from the serum of 297 patients (cancer 81, benign 216) of two hospitals. In this study, the results showed that selecting 2-3 markers with Genetic Algorithm and categorizing them with LDA shows the closest sensitivity, specificity, and accuracy to those of the results obtained through complete enumerations of the combination of 2-4 markers. © 2013 Springer Science+Business Media Dordrecht(Outside the USA).