CART算法在原发性肝癌放疗后HBV再激活的应用

为了建立乙型肝炎病毒(Hepatitis B virus,HBV)再激活的预测模型,提出CART(classification and regression tree)特征选择方法应用在原发性肝癌患者精确放疗后HBV再激活的危险因素分析中,进而建立基于CART和Bayes算法的HBV再激活预测模型。实验结果显示:CART算法划分了多组具有优秀分类能力的特征节点集(危险因素),尤其当特征节点集为HBV DNA水平、外放边界、放疗总剂量、V20和KPS评分时,在CART和Bayes预测模型中的分类正确性分别为88.51%和86.69%,得到HBV再激活正确性贡献度的排序为KPS评分全肝平均剂量V20放疗总剂量V10;当甲胎蛋白AFP出现时,增加了HBV再激活的预测正确性。

Application of HBV reactivation in primary liver carcinoma after radiotherapy based on CART algorithm

To establish an excellent prediction model for Hepatitis B virus reactivation, the CART (classification and regression tree) feature selection method was applied to analyze the risk factors of Hepatitis B virus(HBV) reactivation in patients with primary liver cancer after precise radiotherapy, and then a prediction model of HBV reactivation was established based on CART and Bayes algorithm. The experimental results show that the CART algorithm split multiple sets of feature nodes(risk factors) with excellent classification ability. Especially when the feature set of nodes includes HBV DNA level, outer margin of radiotherapy, the total dose of radiotherapy, V20 and KPS score, the classification accuracy of CART and Bayes prediction models was 88.51% and 86.69% respectively. The decreasing order of accuracy contribution of HBV reactivation was: KPS score,mean dose of liver,V20, the total dose of radiotherapy and V10. The predictive accuracy of HBV reactivation was increased when the alpha-fetoprotein AFP appeared.