Improved Multifactor Dimensionality Reduction for Epistasis Detection

Epistasis detection facilitates determining susceptibility to disease. Multifactor dimensionality reduction (MDR) and multiobjective MDR (MOMDR) were proposed for epistasis detection. However, more measures must be investigated for MOMDR. In this study, we incorporated the Youden index (YI) and correct classification rate (CCR) into MOMDR (MOMDR-YC) for epistasis detection. Simulations were conducted to compare MDR-based YI (MDR-Y), MDR-based CCR (MDR-C), and MOMDR-YC. Moreover, the detection success rates of the three approaches are presented. MOMDR-YC revealed that the YI and CCR measures can enhance the detection success rates of MDR. The simulation results revealed that epistasis could be successfully detected by incorporating YI and CCR into MOMDR.