Dear colleagues,
I would like to make a mathematical equation of Cox regression model, however I am stuck with baseline hazard.
I clearly understand interpretation of coefficients, but I can't understand there can I get baseline risk.
In case of logistic regression, "equivalent" is "constant", which you begin equation with.
But Cox regression doesn't have one.
I'm using SPSS.
Thank you for reply.
#SPSS
#Cox_regression
In the Cox model, the hazard is a high-dimensional nuisance parameter. So it's taken as a matter of fact, not modelled as a constant. This is why the Cox model is sometimes termed semi-parametric, as the hazard is calculated from the data and the coefficients are estimated based on the proportional hazards assumption.
Patrick Royston has done some interesting work on estimating the baseline hazard function using fractional polynomials, which is in the Stata Journal. https://www.stata-journal.com/article.html?article=st0001
Thank you for your answer Ronán.
So if I just want to make an equation with given coefficient, I may not include baseline risk variable if it is not known, and only include significant IV's? And how can I then estimate risk at some time point, e.g. 3 or 6 months?
Thank you
To estimate a risk at a given time point, you need to parametrise the hazard function. Classic choices are AFT models, but have a look at Pat Royston's paper, which offers a more flexible approach.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS