stpm2
stpm2 fits flexible parametric survival models. These models use splines to model some transformation of the survial function. The most common is the $\log[-\log[S(t)]]$ link function, which fits proportional hazards models.
I have added some examples and aim to add to these.
Proportional hazards models
- Comparison with a Cox model
- Simple simulation study to show agreement with Cox model.
- Predicting hazard and survival functions (use of the
timevar()option) - Sensitivity analysis for the number of knots.
- The default knot positions - are they sensible?
- Out of sample predictions (by Sarah Booth)
Time-dependent effects (non proportional hazards)
- Non-proportional hazards
Prognostic Models
Relative survival
- A simple relative survival model
Alternative link functions
- the logistic, probit and Aranda-Ordaz link functions.
