Multistate recursively imputed survival trees for time-to-event data analysis: an application to AIDS and mortality post-HIV infection data.

摘要: This study aimed to introduce recursively imputed survival trees into multistate models (MSRIST) analyze these types of data and identify the prognostic factors influencing disease progression in patients with intermediate events. The proposed method is fully nonparametric can be used for estimating transition probabilities. A general algorithm was provided analyzing multi-state a focus on illness-death progressive models. model considered both beyond Markov Non-Markov settings. We also random (MSRSF) compared their performance classical Cox model. applied dataset related HIV/AIDS based retrospective cohort extracted Tehran from April 2004 March 2014 consist 2473 HIV-infected patients. results showed that MSRIST outperformed using Model MSRSF terms integrated Brier score concordance index over 500 repetitions. identified set important risk as well interactions different states HIV AIDS progression. There are strategies modelling event. adapted two newly developed mining technique (RSF RIST) (MSRSF MSRIST) stages diseases. methods capture any complex relationship between variables useful tool identifying this disease.