How to understand Aalen additive regression model? How to understand these plots with fits from an Aalen’s additive regression model for censored data?

I simply wonder why the curves change between plus and minus. Do they have to understand such that the y-axis is giving the influence in general and the behaviour over time how these specific components contribute?
Precisely: celltypeadeno would be the most significant part and celltypesmallcell, on the other hand, just reduces the other parts?
 A: The Aalen additive model that you show estimates, as functions of time, both the baseline hazard (Intercept) at reference values of covariates and the additive contributions of differences of covariate values from reference values to the hazard of an event. The covariate values can be either fixed in time or time-varying. If the model estimates that the contribution of a covariate to hazard changes from positive to negative over time, then "the curves change between plus and minus."
You'll see that most of "the curves [that] change between plus and minus" have wide ribbons, presumably confidence intervals, covering values of 0. Those probably represent nothing beyond noise for a covariate that has little association with outcome.
The coefficients for celltypeadeno and celltypesmallcell are differences from the reference level of celltype (celltypesquamous for this data set). As the Intercept is the hazard over time at reference levels of all covariates, the Intercept includes patients with squamous-cell tumors.
It's hard to tell what's going on with celltypeadeno from these plots, as its y-axis scale seems to be determined by a very large value at very late times that might not be representative. It's possible that celltypesmallcell is more strongly associated with outcome at early times; you'd have to expand the y-axis scale for celltypeadeno to compare.
