Poisson分布下基于鞍点逼近的慢性病风险差的置信区间构造
发布时间：2018-12-20 13:00
【摘要】：风险差是流行病学中重要的指标之一,常用来比较两种治疗或两种诊断的有效性.因此,风险差区间的精确估计对流行病病情的诊断以及治疗方案的选择有很重要的意义.结合Poisson抽样的优点以及慢性病发病周期长和发病率低的特点,利用鞍点逼近方法来构造了Poisson分布下风险差的置信区间.同时,通过实例和Monte Carlo模拟对传统的四种区间构造方法进行评价.模拟结果表明:在小样本情况下,鞍点逼近方法得到的置信区间大多数能保证覆盖率近似于期望的置信水平并且使得区间长度最短,是一种很好的置信区间构造方法.
[Abstract]:Risk difference is one of the most important indicators in epidemiology and is often used to compare the effectiveness of two treatments or two diagnoses. Therefore, the accurate estimation of the interval of risk difference is very important for the diagnosis of epidemic disease and the choice of treatment scheme. Combined with the advantages of Poisson sampling and the characteristics of long period and low incidence of chronic diseases, the confidence interval of risk difference under Poisson distribution is constructed by using saddle point approximation method. At the same time, the traditional four interval construction methods are evaluated by examples and Monte Carlo simulation. The simulation results show that most of the confidence intervals obtained by saddle point approximation method are similar to the expected confidence level and the interval length is the shortest in the case of small samples. It is a good confidence interval construction method.
【作者单位】： 兰州财经大学统计学院;中国人民大学应用统计科学研究中心中国人民大学统计学院;
【基金】：教育部哲学社会科学研究重大课题攻关项目(15JZD015);教育部人文社会科学重点研究基地重大项目(15JJD910001) 北京市社会科学基金重大项目(15ZDA17) 国家社会科学基金重点项目(13AZD064) 中央高校建设世界一流大学(学科)和特色发展引导专项资金支持(15XNL008) 全国统计科研计划项目(2016LD03) 兰州财经大学“兴隆学者特聘计划”
【分类号】：O212.1
[Abstract]:Risk difference is one of the most important indicators in epidemiology and is often used to compare the effectiveness of two treatments or two diagnoses. Therefore, the accurate estimation of the interval of risk difference is very important for the diagnosis of epidemic disease and the choice of treatment scheme. Combined with the advantages of Poisson sampling and the characteristics of long period and low incidence of chronic diseases, the confidence interval of risk difference under Poisson distribution is constructed by using saddle point approximation method. At the same time, the traditional four interval construction methods are evaluated by examples and Monte Carlo simulation. The simulation results show that most of the confidence intervals obtained by saddle point approximation method are similar to the expected confidence level and the interval length is the shortest in the case of small samples. It is a good confidence interval construction method.
【作者单位】： 兰州财经大学统计学院;中国人民大学应用统计科学研究中心中国人民大学统计学院;
【基金】：教育部哲学社会科学研究重大课题攻关项目(15JZD015);教育部人文社会科学重点研究基地重大项目(15JJD910001) 北京市社会科学基金重大项目(15ZDA17) 国家社会科学基金重点项目(13AZD064) 中央高校建设世界一流大学(学科)和特色发展引导专项资金支持(15XNL008) 全国统计科研计划项目(2016LD03) 兰州财经大学“兴隆学者特聘计划”
【分类号】：O212.1
【相似文献】
相关期刊论文 前10条
1 胡祥;吴涛;李健平;;Poisson分布的参数函数无偏估计[J];重庆工商大学学报(自然科学版);2012年07期
2 范洪福;;一类Poisson分布的数学模型[J];大学数学;2011年04期
3 胡月;多项分布与多元Poisson分布[J];浙江科技学院学报;2005年03期
4 王亮,孙绍荣,李秀森,李世s，
本文编号：2388031
本文链接：https://www.wllwen.com/kejilunwen/yysx/2388031.html
