A flexible semi-Markov model for interval-censored data and goodness-of-fit testing

Yohann Foucher, M. Giral, J.P. Soulillou, and J.P. Daures have a paper to appear in Statistical Methods in Medical Research. They fit a semi-Markov model to a progressive multi-state model with two possible absorbing states. The data are interval censored and the likelihood is calculated through Gauss-Legendre quadrature. A goodness-of-fit procedure, based on comparing observed and expected transitions into the absorbing state is also proposed.

Analysis of interval-censored data from clustered multistate processes: application to joint damage in psoriatic arthritis

Rinku Sutradhar and Richard J. Cook have a new paper in JRSS C. This builds on previous work on random effects multi-state models by Cook, Yi, Lee and Gladman (Biometrics, 2004). Here rather than using a discrete random effects distribution and assuming time homogeneity, they use an MCEM algorithm to allow multivariate normal random effects. In addition, piecewise constant transition intensities allow the assumption of time homogeneity to be relaxed.