SurvivalCurve

survival.SurvivalCurve(
    cumulative_hazard,
    inverse_cumulative_hazard,
    label='curve',
)

A survival curve as its cumulative hazard function, plus how to sample from it.

The families below (exponential, weibull, gompertz), the curve algebra in heormodel.survival.algebra, and heormodel.survival.lifelines_adapter.from_lifelines are the ways to build one; nothing about the analysis code constructs a SurvivalCurve directly.

Parameters

Name	Type	Description	Default
cumulative_hazard	Callable[[_FloatArray], _FloatArray]	The curve’s cumulative hazard H(t), mapping an array of non-negative times to H at each one. survival is exp(-H(t)).	required
inverse_cumulative_hazard	Callable[[_FloatArray], _FloatArray]	The inverse of cumulative_hazard, mapping a target hazard value to the time at which the curve reaches it. sample_time uses it for inverse-transform sampling.	required
label	str	A short description of the curve for repr and for reading back a model built from several curves, for example "weibull(shape=1.2, scale=6)".	'curve'

Example

from heormodel.survival import weibull curve = weibull(shape=1.2, scale=6.0) round(float(curve.survival(6.0)), 5) 0.36788

Methods

Name	Description
sample_time	Sample event times by inverse-transform sampling.
survival	Survival probability S(t) = exp(-cumulative_hazard(t)).

sample_time

survival.SurvivalCurve.sample_time(rng, size)

Sample event times by inverse-transform sampling.

The cumulative hazard evaluated at a curve’s own event time follows a standard exponential distribution, so drawing an exponential variate and inverting the cumulative hazard samples an event time directly, without inverting the survival function on the probability scale.

Parameters

Name	Type	Description	Default
rng	np.random.Generator	A numpy random generator, typically the one an engine’s model function receives.	required
size	int	Number of event times to draw.	required

Returns

Name	Type	Description
	_FloatArray	Array of size sampled event times.

Example

import numpy as np from heormodel.survival import exponential curve = exponential(rate=0.2) rng = np.random.default_rng(0) times = curve.sample_time(rng, 5) times.shape (5,)

survival

survival.SurvivalCurve.survival(time)

Survival probability S(t) = exp(-cumulative_hazard(t)).

Parameters

Name	Type	Description	Default
time	_FloatArray | float	Time (years), scalar or array.	required

Returns

Name	Type	Description
	_FloatArray	Survival probability at each time.

Example

from heormodel.survival import exponential curve = exponential(rate=0.2) round(float(curve.survival(5.0)), 5) 0.36788