Classification System for Multi-class Biomedical Data that Allows Different Data Fusion Strategies

Abstract

Currently, development in high-throughput technologies generate large amount of molecular biology data at awesome rate. How to merge the mass amount of data coming from different sources to obtain significant and complementary high-level knowledge is a state-of-the-art interest in statistics, machine learning and bioinformatics communities. In this article we compare two different data fusion strategies in the context of biomedical data classification using a novel graphical interface of tool for a large-scale data classification system called SPICY. Our classification system allows to compare in controlled environment different fusion strategies for multiple feature selection methods. The results show that properly chosen fusion strategy increases the overall accuracy rate in all tested cases independently of used selection method.