A multi-state model with a random-effects generator matrix

Ardo van den Hout, University College London

Co-authors: Owen Nicholas, University College London

Abstract: A new way of modelling continuous-time multi-state processes is presented by defining the generator matrix by random effects. An application with clinical-trial data illustrates that this makes it possible to define parsimonious models for processes with many states, where the random effects describe the variation across transition intensities.