Figure 4 | Scientific Reports

Figure 4

From: Application of regularized regression to identify novel predictors of mortality in a cohort of hemodialysis patients

Figure 4

AST spline functions. Spline functions to ln transformed serum AST values were fit in a Cox regression for all-cause, cardiovascular (CV), infection-associated (INF), and other mortality causes. Horizontal boxplots represent the distribution of ln-transformed AST values in ISAR patients. Red lines represent the estimated (cause-specific) HR (patients with median AST as reference) and the dashed lines represent the 95% confidence interval of HR at a given AST level. The models were univariate (a) or were adjusted for other relevant predictors in a multivariable model (b) and were fit to the total cohort with non-missing values for the respective variables. Multivariable model predictors are described in “Methods” section. Vertical dashed line represents the cutoff value for the lowest AST quintile.

Back to article page