Pablo Martínez Camblor

Kaplan-Meier estimator in competing risk contexts

Yayın Aralığı: 6
Başlangıç: 2002
Yayıncı: Hacettepe Üniversitesi Fen Fakultesi

Survival analysis has become in a common procedure in biomedical re-searches. Conventionally, the well-known nonparametric Kaplan-Meier(KM) estimator is used in order to approximate the real survivor curve.However, in competing risk contexts where more than one failure causecompete to occur and only one of them is of interest, the direct useof the Kaplan-Meier statistic does not perform correctly and, in or-der to obtain a good estimation, it must be adapted. In this work,via Monte Carlo simulations, the author explores the behavior of theKaplan-Meier estimator in a competing risk context. In addition, dif-ferences between KM and multiple decrement methods are pointed out.Finally, a real-data problem is used in order to illustrate the situation.

Keywords: