Survival analysis is a way of analyzing the expected duration of time until one event occurs (usually death or failure).
In this framework, there are several concepts to take into account:
- An **event** is the experience of interest at hand (death, mechanical failure, recovery, etc) and should be modeled as happening or not.
- **Time** must be defined from the beginning of an observation period until event, end of study or loss of contact.
- **Right-censored data** implies that some of our cannot be considered “over”: the subject may be part of the study or waiting for an event, or the event is not expected to happen anymore (because of loss of contact).
- The **survival function**, $S(t)$, which is by default only dependent on time but that other methods extend to include different covariates.