基于ElasticNet和贝叶斯网络模型的肝硬化并发肝性脑病相关因素分析  

Related factors of liver cirrhosis complicated with hepatic encephalopathy based on Elastic Net and Bayesian network model

作　　者：王旭春 宋伟梅 翟梦梦 任浩 李美晨 全帝臣 张岩波[1] 刘近春[2] 仇丽霞[1] WANG Xu-chun;SONG Wei-mei;ZHAI Meng-meng;REN Hao;LI Mei-chen;QUAN Di-chen;ZHANG Yan-bo;LIU Jin-chun;QIU Li-xia(Shanxi Medical University,Taiyuan,Shanxi 030001,China;不详)

机构地区：[1]山西医科大学公共卫生学院,太原030001 [2]山西医科大学第一医院消化内科

出　　处：《现代预防医学》2021年第9期1705-1709,共5页Modern Preventive Medicine

基　　金：国家自然科学基金面上项目(81973155)。

摘　　要：目的构建肝硬化并发肝性脑病相关因素的贝叶斯网络模型,探索肝硬化并发肝性脑病及其相关因素间的网络关系,通过网络推理反映各因素对肝硬化并发肝性脑病的影响强度。方法采用ElasticNet方法对某三甲医院消化内科950例肝硬化住院患者的临床数据集进行变量初筛,再以MMHC混合算法构建贝叶斯网络,参数估计采用极大似然估计。结果经ElasticNet特征筛选后11个变量最终进入模型;MMHC算法所构建的肝硬化并发肝性脑病贝叶斯网络模型显示:肝肾综合征、电解质紊乱和精神萎靡与HE直接相关,其中肝肾综合征和电解质紊乱是肝性脑病的父节点,精神萎靡为肝性脑病的子节点,即肝肾综合征和电解质紊乱与肝性脑病的发生有关,而肝性脑病会影响肝硬化患者的精神状态;感染是肝性脑病的间接相关因素,通过影响肝肾功能间接诱发肝性脑病。结论贝叶斯网络能很好的揭示肝硬化并发肝性脑病及其相关因素之间的复杂网络联系,进而有针对性的对肝硬化并发肝性脑病进行预防控制,在分析相关因素上具有更好的应用前景。Objective To establish a Bayesian network model of factors related to liver cirrhosis complicated with hepatic encephalopathy,and to explore the network relationship between liver cirrhosis complicated with hepatic encephalopathy and its related factors,so as to reflect the influence strength of influencing factors on liver cirrhosis complicated with hepatic encephalopathy through network reasoning.Methods The ElasticNet method was used to conduct variable preliminary screening on the clinical data set of 950 inpatients with liver cirrhosis in the Gastroenterology Department of a tertiary hospital.Then the Bayesian network was constructed with MMHC hybrid algorithm and maximum likelihood estimation was used for parameter estimation.Results Eleven variables finally entered the model after being screened by the ElasticNet regression features.The Bayesian network model of liver cirrhosis complicated by hepatic encephalopathy constructed by the MMHC algorithm showed:Hepatorenal syndrome,electrolyte disorder and malaise were directly related to hepatic encephalopathy,of which hepatorenal syndrome and electrolyte disorder were the parent nodes of hepatic encephalopathy,and malaise was the child nodes of hepatic encephalopathy,which meantthat hepatorenal syndrome and electrolyte disturbance were related to the occurrence of hepatic encephalopathy,and hepatic encephalopathy could affect the mental state of patients with cirrhosis.Infectionwas an indirect related factor of hepatic encephalopathy,which indirectly induced hepatic encephalopathy by affecting liver and kidney function.Conclusion Bayesian network can well present the complex network connection between liver cirrhosis complicated by hepatic encephalopathy and its related factors,and then conduct targeted prevention and control of liver cirrhosis complicated by hepatic encephalopathy,with better analysis of related factors application prospects.

关 键 词：肝性脑病 ElasticNet MMHC算法 贝叶斯网络 

分 类 号：R575.3[医药卫生—消化系统]