Specifically, the data set consists of a heterogeneous group of individuals which can be divided into many sub-groups. The random effects (e.g. intercepts and/or slopes) in a mixed COX model could refer to these sub-groups. IN addition, there are several other potential predictor variables. As always, I would be grateful for input from anyone interested.
As preliminary (non-expert) thoughts, it seems that Cox regression is often used because it is convenient and familiar. However, it imposes parametric constraints on the underlying distributions (for example, even the random effects often are assumed to be normally distributed). Random forests do not rely on the same assumptions and may perform equally well. Is this interpretation reasonable? What are the disadvantages of using random forests for survival analysis?
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- DNA