Advances and Applications in Statistics
Volume 41, Issue 2, Pages 137 - 165
(August 2014)
|
|
LONGITUDINAL DATA ANALYSIS STRATEGIES - AN APPLICATION OF MIXED MODEL TO POST - TRANSPLANT SERUM CREATININE DATA
Zailong Wang
|
Abstract: In transplant clinical studies, serum creatinine data are collected repeatedly on same individuals over time to monitor individual renal function changes and to determine treatment effects on renal function as well as on other efficacy and safety endpoints. Hence longitudinal analysis methods are appropriate to model such data. In this paper, we systematically present the strategies of using mixed-effect regression model to analyze serum creatinine data from a Novartis transplant clinical trial. These strategies include fixed-effect covariate selection approach; covariance model selection; and model based treatment effect tests. Furthermore, various types of residuals from mixed-effect regression model and influence diagnostics are discussed in detail for model diagnosis. Related background of longitudinal analysis and clinical trial is also presented. |
Keywords and phrases: mixed-effect regression, covariance structure analysis, residual analysis, serum creatinine data. |
|
Number of Downloads: 357 | Number of Views: 1044 |
|