Identification of Significant Gene Expression in Liver Cancer-Induced HBx Virus Using Enhanced Machine Learning Method

Hepatitis B virus (HBV) is one of the causes to induce liver chronic disease until liver cancer. The virus is inserted and integrated through the host’s DNA. It affected cell cycles improperly. Microarray technology is a tool to investigate gene expression by quantifying RNA in the liver cancer mechanism. However, the number of genes involved is very huge, then it is required to know the potential gene for the classification task in the liver cancer mechanism. Therefore, this paper aims to purpose the gain-ratio measurement to select the significant feature as a predictor in the classifier model. The feature selection is a method based on the entropy value to select the significant gene expression. The selected gene is used to build the classifier model of the representative machine learning algorithms including SVM, Naïve Bayes, KNN, C5.0 Decision Tree, and Random Forest. The experimental results show that the performance result including the accuracy, sensitivity, specificity, and AUC are high. Also, the time computation of the algorithms using feature selection is much shorter than without using feature selection for prediction in the liver cancer mechanism.