Penalised maximum likelihood estimation in multi-state models for interval-censored data

Continuous-time multi-state Markov models can be used to describe transitions over time across health states. Given longitudinal interval-censored data on between states, statistical inference changing is possible by specifying for transition hazards. Parametric time-dependent hazards restrictive, and nonparametric hazard specifications using splines are presented as an alternative. The smoothing of the controlled penalised maximum likelihood estimation. With multiple in a model, there penalty parameters selecting optimal amount challenge. A grid search estimate computational intensive especially when combined with methods deal times. new efficient method proposed where estimation automatic. simulation study undertaken validate illustrate effect interval censoring. feasibility illustrated two applications.