项目名称: 生物医学研究中不完全分类数据的统计推断
项目编号: No.11201412
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 数理科学和化学
项目作者: 李会琼
作者单位: 云南大学
项目金额: 23万元
中文摘要: 在生物医学研究中经常需处理分类数据,分类数据分析是目前国内外研究的热门课题之一。然而,在两种处理方法的配对研究中,由于被研究的配对对象的去死、搬迁或处理方法本身的副作用等原因致使研究人员不能得到被试者的数据,因而造成一些数据的缺失;在纵向数据研究中,由于被调查者的出国或工作调动、或者死亡等原因而导致数据的缺失,这就是所谓的不完全数据。在这种情况下,标准的统计方法不能直接应用到这些不完全分类数据的统计分析。为此,本项目将基于不完全2×联表、高维的不完全2×联表及不完全的有序分类数据构成的K×研究完全可观测的配对数n和不完全可观测的配对数m是随机变量时的有关统计推断问题,建立一套系统分析不完全分类数据的分析理论,探索出计算置信区间上下限的简单易行的方法,同时将带有MAR缺失数据机制的各种统计方法推广到不可忽略缺失机制下的不完全分类数据中。
中文关键词: 不完全数据;置信区间;列联表;贝叶斯分析;
英文摘要: In biomedical research, ones often need to deal with categorical data, which is becoming one of the hot issues to be discussed in the present domestic and foreign statistical analysis. However, in the matched-pair study for two kinds of treatments, the data is missing because of death、migration of matchable subjects or drop out due to adverse effect caused by treatments;In longitudinal data,because the respondents are abroad or transfer, or died, so that researchers can not completely observe the data of subjects, this is the so-called incomplete data. In this case, the standard statistical methods cannot be applied directly to analy these incomplete categorical data. Based on the incomplete 2×contingency tables, high dimensional incomplete 2×contingency tables and incomplete ordered categorical data form the K×table,when n and m are random,this project will study their statistical inference. This project will also provide a set of systematic analysis theory of the incomplete data, explores the simple and easy method to calculate confidence intervals. Meanwhile, various statistical methods with the MAR missing data mechanism apply to nonignorable mechanism.
英文关键词: incomplete datas;confidence intervals;contingency tables;bayes analysis;