Abstract | Keywords | Export | Availability | Bookmark

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Chronical kidney disease (CKD) is considered to be an important problem in public health nowadays. Its prevalence is increasing worldwide. CKD can lead to severe complications that require costly therapy. In the first stages, however, the disease is typically progressing without any specific symptoms. Knowledge of the associations between comorbidities can help detect CKD earlier and is therefore of primary importance. Of particular interest is the relationship between kidney function and blood pressure. On the one hand, hypertension is usually seen as both a cause and a consequence of the CKD. On the other hand, some studies have claimed that the decrease blood pressure leads to decrease in kidney function in older patients. In this study, three methods from statistical analysis and data mining are applied to the longitudinal data set from Intego database in order to examine the evolution of kidney function with blood pressure in patients older than 40 years old. The techniques under consideration are namely linear mixed-effects regression modelling, regression trees and a combination of them, RE-EM tree. The advantages and shortcomings of the methods are discussed. The resulting models are compared. It is demonstrated that linear mixed-effects regression model describes the data better than tree based predictors, which indicates a strong linear relationship between eGFR and covariates under consideration. The biggest amount of change in kidney function is concluded to be due to the aging process.