Multi-State Models with Practical Applications (303114)

Keywords: Biostatistics, Multi-State Models, Statistical Software

Multi-state models (MSMs) are natural and powerful modeling approaches in time-to-event problems. From an inferential viewpoint, MSMs rely on the estimation of hazard functions that characterize the instantaneous probability of transitioning from one state to another at any given time. The simplest MSM is the “Illness-Death Model” that has been successfully applied in numerous areas of research, including epidemiology, biostatistics and health care economics. Due in part to software availability, in the recent years MSMs have been applied to increasingly complex biomedical studies, such as organ transplantation with semi-competing risks, cancer studies in which multiple treatment regimens are offered and repeated hospitalization studies. Despite their usefulness, MSMs have yet to become mainstream in the applied statistical community at-large. We aim to increase awareness of the potential advantages of MSMs and to present relevant data examples. To visualize the MSM and conduct analyses, several R packages, including diagram, etm, mstate, msm, mvna, survival and timereg will be discussed, along with relevant data examples.