Nonparametric Estimation in Interval-Censored Multi-State Processes

L. F. Meira Machado

Multi-state models are widely used to describe the evolution of individuals across states over time, but in many studies event times are only known within inspection intervals, leading to interval-censored data. We propose a flexible nonparametric framework for estimating state occupation and transition probabilities in progressive multi-state models under interval censoring. The approach represents occupation probabilities through latent transition indicators, replaced by conditional probabilities defined over Turnbull intervals. We also address the estimation of transition probabilities by adapting Aalen–Johansen-type constructions to interval-censored data. The framework accommodates settings with partially or fully interval-censored transitions and provides practical tools for inference.

Keywords: Interval censoring, Multi-state models, State occupation probabilities, Transition probabilities, Turnbull estimator

SI: Portuguese Statistical Society invited session in Statistics and Probability
September 2, 2026  11:20 AM
Aula 29

N. M. Brites, J. Brazão, M. Reis

R. Gaio, R. Costa-Miranda, W. González-Manteiga

C. Tenreiro