Home › Survival › Survival estimators

KaplanMeierCurve

A Kaplan-Meier survival curve with its Greenwood variance and interval.

public sealed record KaplanMeierCurve(SurvivalStep[] Steps, double[] Survival, double[] Lower, double[] Upper, double ConfidenceLevel)

Parameters — Steps are the curve's steps, ascending in time and starting at zero. Survival, Lower and Upper hold the estimate and its two bounds, one entry per step. ConfidenceLevel is the level the bounds were built at.

Example — the four arrays share one index.

using Lodestar.Survival;

KaplanMeierCurve curve = KaplanMeier.Estimate([1, 2, 3], [true, true, true]);

int steps = curve.Steps.Length;  // => 4
double level = curve.ConfidenceLevel;  // => 0.95
double lastSurvival = curve.Survival[3];  // => 0

Remarks — the bounds are built on the log-log transform of the estimate, which is lifelines' default. They are therefore not symmetric about Survival, and they are not the estimate plus or minus its own standard error — that interval can leave [0, 1], and at S = 0.857 on 21 subjects it reaches 1.0067.

Where the curve reaches zero, as above, the transform is undefined and both bounds collapse onto the estimate.

The four arrays are taken and exposed as they are, not copied: writing to one changes this record and what it equals, and every holder of the same array, a with copy included (#1305).

Applies to — net10.0, netstandard2.0.

See also — KaplanMeier.Estimate, SurvivalStep.

Members

member	what it does
KaplanMeierCurve.Equals	Value equality over the steps and the three curves.
KaplanMeierCurve.GetHashCode	A hash consistent with it.