基于遗传算法特征选择的HBV再激活分类预测模型

探讨原发性肝癌患者精确放疗后乙型肝炎病毒(hepatitis b virus,HBV)再激活的危险特征和分类预测模型。提出基于遗传算法的特征选择方法,从原发性肝癌数据的初始特征集中选择HBV再激活的最优特征子集。建立贝叶斯和支持向量机的HBV再激活分类预测模型,并预测最优特征子集和初始特征集的分类性能。实验结果表明,基于遗传算法的特征选择提高了HBV再激活分类性能,最优特征子集的分类性能明显优于初始特征子集的分类性能。影响HBV再激活的最优特征子集包括:HBV DNA水平,肿瘤分期TNM,Child-Pugh,外放边界和全肝最大剂量。贝叶斯的分类准确性最高可达82.89%,支持向量机的分类准确性最高可达83.34%。

HBV reactivation classification prediction model based on feature selection of genetic algorithm

This study investigates the risk features and classification prognosis models for hepatitis b virus (HBV) reactivation in patients with primary liver carcinoma after precise radiotherapy (RT). Feature selection method based on Genetic Algorithm (GA) is proposed, the optimal feature subsets are selected from initial feature sets of primary liver carcinoma. HBV reactivation classification prediction models of Bayes and support vector machine (SVM) are built, and the models are used to evaluate predict the classification performance of the optimal feature subsets and initial feature sets. The experimental results show that feature selection based on GA improved the classification performance of HBV reactivation, and the classification performance of the optimal feature subset is much better than the initial features set. The optimal feature subset affecting HBV reactivation include HBV DNA level, tumor staging TNM, Child-Pugh, outer margin of RT and maximum dose of liver. The classification accuracy of Bayes is up to 82.89%, and the classification accuracy of SVM is up to 83.34%.