Mixed Response Model of Binary Response of Fatty Liver and Dyslipidemia and Its Application
In order to explore the relationship between fatty liver,dyslipidemia and various routine physical examination items,and achieve risk assessment of fatty liver and dyslipidemia,a binary response generalized linear mixed model with dyslipidemia as the response variable was established based on the results of 9091 blood routine physical examinations of 2972 examinees.Disease risk prediction and model prediction effectiveness evaluation were conducted.The results show that the common influencing fac-tors of dyslipidemia in fatty liver include gender age,diabetes,body mass index,alanine aminotransferase and high-density lipo-protein,the mixed effects of fatty liver and dyslipidemia responses show a certain positive correlation.In the model prediction,the area under the ROC curve SAUC of fatty liver was 0.855(95％CI:0.831～0.881),and the area under the ROC curve SAUC of dyslip-idemia was 0.814(95％CI:0.794～0.850),five cross validation results show that the SAUC of lipid abnormalities in fatty liver were 0.850(95％CI:0.845～0.868)and 0.807(95％CI:0.793～0.819),respectively.The dual response model can fully reflect the corre-lation between response variables and has good predictive ability,making it suitable for joint analysis of multiple medical diseases.
physical examination datafatty liverdyslipidemiaroutine physical examination itemsmultivariate generalized linear mixed effect model
万方数据
维普
