首页 | 本学科首页   官方微博 | 高级检索  
检索        

医学科研中缺失数据的EM估计
引用本文:陈长生,王彤,徐勇勇,尚磊.医学科研中缺失数据的EM估计[J].第四军医大学学报,2002,23(1):59-61.
作者姓名:陈长生  王彤  徐勇勇  尚磊
作者单位:1. 第四军医大学预防医学系卫生统计学教研室,陕西,西安,710033
2. 山西医科大学公共卫生学院卫生统计学教研室,山西,太原,030001
基金项目:国家自然科学基金资助项目 (3 990 0 12 6)
摘    要:目的 探讨一种数据补缺的统计方法,以避免医学科研中常常发生观测数据的缺失而造成统计分析的困难。方法 通过数据间的相关信息,应用EM算法和生长曲线模型进行数据分析。结果 给出了EM算法的迭代步骤,并用软件SAS/IML编程实现。结论 通过EM补缺,可以保证资料分析的有效性,且EM算法补缺使参数估计的方差大大变小,从而提高了参数估计精度。

关 键 词:EM算法  缺失数据  重复观测  数据分析  医学科研
文章编号:1000-2790(2002)01-0059-03
修稿时间:2001年10月5日

EM estimation for missing data in medical research
CHEN Chang Sheng ,WANG Tong ,XU Yong Yong ,SHANG Lei.EM estimation for missing data in medical research[J].Journal of the Fourth Military Medical University,2002,23(1):59-61.
Authors:CHEN Chang Sheng  WANG Tong  XU Yong Yong  SHANG Lei
Institution:CHEN Chang Sheng 1,WANG Tong 2,XU Yong Yong 1,SHANG Lei 1 1Department of Health Statistics,Faculty of Preventive Medicine,Fourth Military Medical University,Xi'an 710033,China,2Department of Health Statistics,Faculty of Pub
Abstract:AIM Missing data often arised in medical researches, the difficulties of statistical analysis existed owing to missing data. This paper explored a kind of statistical method used for tackling missing data. METHODS Based on correlation information among data, the authors analyzed data by using EM algorithm and growth curve model. RESULTS The iterative steps of EM algorithm were given, and a proper program was implemented in SAS/IML code. CONCLUSION The validity of analyzing data imputed by EM would be ensured, and the variances of parameters estimation would be greatly reduced, so accuracy of parameters estimation would be improved.
Keywords:EM  algorithm  missing  data  repeated  measures  data  analysis
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号