Proportional Hazards Assumption Test and Nonlinear Relationship Analysis of the Multivariable Cox Regression Model

A: Schoenfeld residual plot for the multivariable Cox proportional hazards regression model to test the proportional hazards (PH) assumption. The plot shows the relationship between residuals of each covariate and time. A smooth curve that is approximately horizontal and residuals that are randomly distributed around zero provide evidence that the PH assumption holds for that variable. B: Restricted cubic spline (RCS) analysis of the multivariable Cox model was used to explore the ​​nonlinear relationship​​ between a continuous covariate and the Llog hazard ratio. The solid line represents the fitted association curve, and the shaded area represents the 95% confidence interval.