Study On The Classification Prognosis Model Of HBV Reactivation Based On Intelligent Computing

Primary liver carcinoma patients after precise radiotherapy is easily lead to HBV reactivation.Currently,the risk factors for HBV reactivation after precise radiotherapy in primary liver carcinoma patients need to be studied,and intelligent prognosis model also needs to be established.We proposed the research on classification prognosis model of HBV virus reactivation based on intelligent computing in this paper.The clinical data set of 90 primary liver carcinoma patients after precise radiotherapy was studied.It is difficult to directly identify the risk factors for HBV reactivation and to establish classification prognosis model of HBV reactivation with excellent classification ability.Therefore,we use the feature selection algorithm to select the risk factors set for HBV reactivation,and then establish the HBV reactivation classification prognosis model.This is also the focus of this article.We adopt two kinds of ideas to establish classification prognosis model of HBV reactivation in this paper.The first idea is to use genetic algorithm to find out the risk factors of HBV reactivation from initial clinical data set of primary liver carcinoma,and then to establish classification prognosis model of HBV reactivation based on BP and RBF neural networks.The second idea is use CART algorithm to establish HBV reactivation classification prognosis model.The experimental results showed that genetic algorithm selects "HBV DNA level","outer margion of radiotherapy","tumor staging TNM","KPS score" and "Child-Pugh" have the best classification prognosis performance in BP and RBF neural networks prognosis models,the accuracy is 82.21% and 83.31% especially.The classification accuracy improved 10% and 11.1% than the initial data set.The classification performance of RBF prognosis model is better than BP.The CART algorithm selects the risk factors to construct the nodes of the CART tree from the initial data set.CART has the best classification performance when selecting "HBV DNA level","outer margion of radiotherapy","Total dose of radiotherapy","V20" and "KPS score".The accuracy is 88.51%.CART fully demonstrated the relationship between the risk factors of nodes,more conducive to the understanding for clinicians.The proposed classification prognosis model of HBV reactivation is great significance in guiding radiotherapy for primary liver carcinoma patients.