基于双变量的诊断试验Meta分析

Bivariate Meta-analysis of Diagnostic Tests

为改进双变量诊断试验meta分析模型,提高诊断准确性.在随机效应模型下,对双变量诊断试验meta分析引入组内相关性,构建新模型,并进行统计实现.结果表明:由新模型得到的灵敏度、特异度分别为0.531、0.923,均比原模型下的值增大;诊断比值比为13.551,是原模型分析结果的2倍.证明引入组内相关性后,可以充分利用各研究的原始数据信息,得到稳定可靠的结果,在双变量诊断试验的meta分析中有很好的应用价值.

Our objective was to enhance the diagnostic accuracy by improving the model of bivariate recta- analysis of diagnostic tests. At the background of random effects,a new model was set up,leading the correlation within the groups into the bivariate meta-analysis of diagnostic tests,and implementing it with statistic methods. Results showed that the values of specificity and sensitivity in the new model are bigger than those in the original model,and that the diagnostic odds ratio is also higher than the results of the original one. After adding correlation within the groups to the new model,we can make full use of the whole information of the original data,and obtain stable and reliable results,which have a good value of application in the bivariate meta-analysis of diagnostic tests.