生存分析参数回归模型拟合及其SAS实现

发布时间：2018-01-28 06:10

本文关键词： 生存分析参数回归拟合优度检验预测　出处：《山西医科大学》2008年硕士论文　论文类型：学位论文

【摘要】： 目的:阐述生存分析参数回归模型的拟合、拟合优度检验、生存预测、分段指数回归方法及其SAS实现过程。对生存资料建立参数回归模型,对生存时间的分布特征进行描述,对影响生存时间的主要因素进行分析。在拟合优度检验筛选模型基础上进行生存预测。方法:采用SAS8.2中的PROC LIFEREG过程,首先对糖尿病患者生存资料建立指数模型、Weibull模型和Gamma模型,然后用图示法和似然比检验两种方法对所建模型进行拟合优度检验,比较该资料拟合哪种模型较好。根据筛选模型预测中位生存时间,采用宏程序预测生存率及产生风险函数图。若生存资料不符合指数模型,可考虑建立分段指数回归模型。结果:参数回归模型可分析影响生存时间的协变量,给出协变量系数的参数估计值、标准误和假设检验结果。图示法和似然比统计量两种方法对参数模型进行拟合优度检验各有其优缺点,图示法比较直观,而似然比统计量更加准确。预测中位生存时间、生存率和风险函数可反映疾病对生命的危害程度,用于评价某些病程较长疾病的远期疗效。SAS软件实现非常方便和快捷。结论:在生存分析中用SAS程序对生存资料拟合适当的模型,可以更方便迅速地对具有截尾数据的资料进行参数估计和假设检验。当资料服从特定分布时应选用相应的参数回归模型。图示法和似然比检验可用于参数回归模型的拟合优度检验。中位生存时间、生存率和风险函数是预后评价的主要指标。
[Abstract]:Objective: to describe the regression model of survival analysis parameters, such as fitting, goodness of fit test, survival prediction, piecewise exponential regression method and its SAS implementation process, and to establish a parameter regression model for survival data. The distribution characteristics of survival time are described, the main factors affecting survival time are analyzed, and the survival prediction is carried out on the basis of the selection model of goodness of fit test. Methods: using the PROC LIFEREG process in SAS8.2, an exponential model and a Gamma model were established for the survival data of diabetic patients. Then, the two methods of graph method and likelihood ratio test were used to test the goodness of fit of the model, and the model was compared. The median survival time was predicted according to the screening model. The macro program is used to predict the survival rate and the risk function graph. If the survival data do not conform to the exponential model, a piecewise exponential regression model can be considered. Results: the parameter regression model can analyze the covariables that affect the survival time, and give the parameter estimate of the covariable coefficient. The results of standard error and hypothesis test. The graphical method and the likelihood ratio statistic method have their own advantages and disadvantages to test the goodness of fit of the parameter model, and the graphical method is more intuitive. But likelihood is more accurate than statistics. Predicting median survival time, survival rate and risk function can reflect the degree of disease damage to life. SAS software is very convenient and fast to evaluate the long-term effect of some long-term diseases. Conclusion: the survival data were fitted with SAS program in survival analysis. The parameter estimation and hypothesis test of the data with truncated data can be carried out more conveniently and quickly. The corresponding parameter regression model should be selected when the data is applied from a particular distribution. The graphical method and the likelihood ratio test can be used for the parameter regression model. Test of goodness of fit. Median survival time. Survival rate and risk function are the main indicators of prognosis evaluation.
【学位授予单位】：山西医科大学
【学位级别】：硕士
【学位授予年份】：2008
【分类号】：R311

【共引文献】