节点文献
基于隐马尔科夫模型的重型肝炎演化研究
The Study on The Evolution of Severe Hepatitis Based on The Hidden Markov Model(HMM)
【作者】 张玲;
【导师】 邢永康;
【作者基本信息】 重庆大学 , 计算机软件与理论, 2015, 硕士
【摘要】 在我国,慢性乙型肝炎是一种常见的疾病,因其具有传染性而严重威胁着人们的身体健康。慢加急性肝衰竭(acute-on-chronic liver failure,ACLF)是在慢性肝病的基础上,由某些原因引发急剧肝损伤而导致的急性肝功能失代偿,慢加急性肝衰竭的临床治疗难度大、死亡率高使其成为威胁被乙型肝炎病毒(hepatitis B virus,HBV)感染人群生命健康的重大疾病,也成为研究的热门领域之一。因此,及时选择有效的方法治疗是治愈且降低死亡率的关键。对于普通肝病或慢性肝病患者,如果能及时且准确的评估患者的病情发展趋势,将为肝病患者制定个体差异化的救治方案奠定基础,在保证治疗效果的同时为患者降低治疗费用、减少了痛苦。但是现在关于ACLF的研究大多集中在肝衰竭的预后评估、其它肝病并发症及其防治、发病机制等方向,而ACLF前期临床研究及病情预测方面的研究较少。本文的主要工作是研究重型肝炎中慢加急性肝衰竭的演化过程,通过与第三军医大学第一附属医院西南医院传染病科的合作,通过对病人医学化验数据的处理与分析研究,在以往慢性肝炎和慢加急性肝衰竭预后研究的基础上,提出了根据病情发展的特性确定重型肝炎演化过程中存在着几个隐藏的过程,并根据病人各个时间段的化验指标作为观察状态建立隐马尔科夫模型预测病情的发展,通过实验验证和评价了该预测模型,最后提出重型肝炎演化过程中的关键状态的定义,并通过算法来确定此关键状态。本文针对以上提出的问题进行研究,主要工作如下:①在分析前人关于各项指标对病情影响的研究基础上,提出慢性重型肝炎病程发展过程的预测,该预测模型是关注整个病程的发展过程的预测模型,不为寻找指标间的线性关系;②提出了根据病情发展的特征确定重型肝炎演化过程中存在着几个隐藏的过程,并根据病人各个时间段的化验指标作为观察状态建立隐马尔科夫模型预测病情的发展趋势,通过实验验证预测准确率,并通过ROC分析评价该预测模型;③提出重型肝炎演化过程中关键状态的研究与定义,根据学习得到的隐马尔科夫模型通过算法计算得到关键状态,与西南医院医学研究人员合作,用医学角度解释关键状态存在的意义。实验证明,基于隐马尔科夫模型预测ACLF发展比以往的方法预测准确率要高且ROC曲线下面积AUC较优,重型肝炎演化过程中确实存在关键状态,为重型肝炎的诊断与治疗提供了参考,在医学研究上有重大意义。
【Abstract】 Chronic hepatitis B is a common disease that seriously threatens people’s health. Acute-on-chronic liver failure(ACLF) is the acute decompensated liver function caused by sharp liver injury based on chronic liver disease. ACLF not only has become the serious disease which threatens the life and health of the HBV-infected people as a result of the difficult clinical treatment and the high mortality rate, but also is one of the hot researched fields. Therefore, the timely and effective treatment is vital to reduce the mortality rate. If the developmental trend of a patient with the liver disease, who is just sent to the hospital, can be assessed timely and accurately, an individual therapy would be established, which not only guarantees the therapeutic effect to reduce the pain for the patient but also decreases the cost of treatment. Nowadays the studies on ACLF focus on pathogenesis, complication, remedy and prognostic assessment, etc., while there are few studies on preclinical studies and prognosis of ACLF.This thesis aims to investigate the evolutionary process of the severe hepatitis. Through cooperating with the Department of Infectious Disease in Southwestern Hospital-the First Affiliated Hospital of Third Military Medical University, treating and analyzing patient’s medical laboratory data and on the basis of prognostic study on chronic hepatitis and ACLP, this thesis has put forward to determine several hidden evolutionary processes of severe hepatitis according to the characteristics of the development of the disease. This study also establishes HMM to predict the development of the disease according to the lab indexes of the patient in different periods. Finally, this research has put forward to the definition of the key state which is determined by arithmetic in the evolutionary process of the severe hepatitis by the experimental verification and the assessment of HMM.① In order to solve the above research problems, the main contents of this dissertation include: Putting forward to the prediction of the development process of the disease and the study on the early warning of chronic severe hepatitis.② Determining several hidden evolutionary processes of severe hepatitis according to the characteristics of the development of the disease and establishing HMM to predict the development of the disease according to the lab indexes of the patient in different periods; predicting the accuracy rate by the experimental verification; analyzing and accessing the HMM by ROC.③ Putting forward to the definition and the study of the key state in the evolutionary process of severe hepatitis; gaining the key state according to the HMM and arithmetic and explaining the significance of the key state from the medical standpoint through cooperating with the medical researchers in Southwestern Hospital.The experiment proves that the prediction of the development of ACLF based on HMM is more accurate than that in other ways. What’s more, AUC is much better. This study also found that the key state really exists in the evolutionary of severe hepatitis which provides a reference for the diagnosis and treatment of the severe hepatitis. This study possesses a great significance in medical study.
【Key words】 Hidden Markov model; severe hepatitis; prediction model; the critical state;