Association Analysis Of Binary Variables In Group Testing

Grouping detection means that all samples to be tested are mixed together and divided into several groups.The size of each group is k,and then targeted detection is carried out according to one group as a unit.Grouping detection can significantly reduce the detection cost,and to a certain extent can improve the accuracy and efficiency of detection,so it is widely used in the fields of epidemic diseases,genetic inheritance,blood detection,plant pathology and so on.When grouping a disease for detection,the detected male and female individuals are distributed mixed in each group,in order to determine the relationship between disease distribution and gender,we proposed the likelihood ratio test method.In addition,when test object is normal men and women and ill mixed together,each group of the male individual number for s,values range from 1 to k-1,the number of men,value range is 0 to s,each group of female individual number k-s,number for women,the value range is 0 to k-s,so obey gaussian mixture model was test sample distribution,implicit variable is set in the number and the number of women,men due to the limited parameters,need to estimate the parameters of the seven,based on the limited parameters of the EM algorithm to estimate its parameters,and to estimate the parameters of the generation into the likelihood ratio test statistic,According to the distribution of the test statistic,the rejection domain of likelihood ratio test is deduced.Paper first is deduced under the normal distribution and gaussian mixture distribution model parameter estimation formula,then according to the likelihood ratio is given the likelihood ratio test statistics and their distribution,through the computer simulation method for the inspection and evaluation,the results show that the likelihood ratio test methods in the distributed detection has good statistical properties.