Abstract
In this paper, we consider the joint modelling of survival and longitudinal data with informative observation time points. The survival model and the longitudinal model are linked via random effects, for which no distribution assumption is required under our estimation approach. The estimator is shown to be consistent and asymptotically normal. The proposed estimator and its estimated covariance matrix can be easily calculated. Simulation studies and an application to a primary biliary cirrhosis study are also provided.
| Original language | English |
|---|---|
| Pages (from-to) | 571-589 |
| Number of pages | 19 |
| Journal | Scandinavian Journal of Statistics |
| Volume | 45 |
| Issue number | 3 |
| Early online date | 30 Jan 2018 |
| DOIs | |
| Publication status | Published - Sept 2018 |
Keywords
- Cox model
- informative observation times
- log-normal distribution
- longitudinal data
- multistate models