Abstract:
|
In scientific studies in the engineering and in other settings, the time-to-terminal-event (TTTE) for a unit (e.g., machine, equipment, patient, mouse, cell, subject, medical equipment, etc.) is of interest. The terminal event (TE) could, for instance, be the failure or death of the unit. For each unit there could be other observable processes, such as the occurrences of different types of recurrent events, referred to as recurrent competing risks (RCR), and the trajectory of a longitudinal marker (LM). These recurrent competing risks and longitudinal marker processes may be associated with each other, but more importantly, these processes may impact the occurrence of the terminal event. Therefore, a proper understanding and an effective modeling of the relationships among recurrent competing risks, longitudinal marker, and terminal event, together with covariate processes, offer potential of performing unit-specific interventions to prolong the occurrence of a detrimental terminal event or hasten the occurrence of a beneficial terminal event. We present in this talk joint dynamic models and will describe statistical inference methods and time-to-terminal-event prediction methods.
|