CMStatistics 2016: Start Registration
View Submission - CMStatistics
B0224
Title: Nonparametric estimation of state occupation and transition probabilities in non-Markov multistate models Authors:  Jan Beyersmann - Ulm University Institute of Statistics (Germany) [presenting]
Arthur Allignol - Ulm University (Germany)
Carina Mueller - Metronomia Clinical Research GmbH Munich (Germany)
Abstract: The Aalen-Johansen estimator generalizes the Kaplan-Meier estimator for independently left-truncated and right-censored survival data to estimating the transition probability matrix of a time-inhomogeneous Markov model with finite state space. The Markov assumption enables use of martingale methods to investigate properties of the Aalen-Johansen estimator. It has been noted that the Aalen-Johansen estimator, standardized by a consistent estimator of the initial distribution of the multi-state model, consistently estimates the state occupation probabilities of a non-Markov model if censoring is entirely random. We extend this result to random left-truncation, improve on martingale arguments used to establish these results and discuss how transition probabilities may subsequently be estimated.