math concept 6 topics use this
Math concept
Survival Analysis
Core equation
$$S(t) = P(T > t) = \exp\!\left(-\int_0^t \lambda(u)\,du\right)$$
Survival analysis studies the time until an event — death, failure, relapse. The hazard function, survival function, and censoring are its core concepts, connecting biostatistics, actuarial science, reliability engineering, and economics.

The hazard and survival functions

The hazard rate $\lambda(t) = \lim_{dt\to0} P(T \in [t,t+dt) \mid T \geq t)/dt$ — the instantaneous event rate.

The survival function $S(t) = P(T > t) = \exp(-\Lambda(t))$ where $\Lambda(t) = \int_0^t\lambda(u)\,du$ is the cumulative hazard.

The relationship: $f(t) = \lambda(t)S(t)$, $S(t) = 1 - F(t)$.

Censoring

Right censoring: we observe $\min(T, C)$ where $C$ is a censoring time. The observed data are $(t_i, \delta_i)$ with $\delta_i = \mathbf{1}[T_i \leq C_i]$.

Censoring is crucial: ignoring it (treating censored times as event times) biases estimates downward.

Kaplan-Meier estimator

The non-parametric estimator of $S(t)$:

\[\hat{S}(t) = \prod_{t_i \leq t} \left(1 - \frac{d_i}{n_i}\right)\]

where $d_i$ events occur among $n_i$ at-risk individuals at time $t_i$.

Fields that use this concept
Finance & economics Actuarial science
Life sciences Biostatistics
Appears in fields Actuarial science Biostatistics
Difficulty
intermediate