Survival analysis in living and engineering sciences

Abdoulaye Diamoutene (1, 2) , Diakarya Barro (3) , Serge Manituo Aymar Somda (4) , Farid Noureddine (2) , Bernard Kamsu-Foguem (2)

1 Institut polytechnique rural de formation et de recherche appliquée de Katibougou
2 LGP - Laboratoire Génie de Production
3 Université de Ouaga II
4 Centre Muraz [Bobo-Dioulasso, Burkina Faso]

Survival or reliability analysis is one of the most significant advancements of statistics in the last quarter of the 20th century. This domain of statistics takes an important place in biomedical and industrial framework. In this paper, we propose a new baseline hazard function from extreme value theory. The newly suggested function is non-monotone and is named as a generalized extreme values baseline hazard function. We prove that this function satisfies hazard properties. A study of the characteristics of the function related to time is made. Conditions for applicability of the model are obtained.

Mots clés: Survival analysis; Reliability; Hazard function; Parametric proportional; Hazard model; Extreme value theory

Domaines: Modélisation et simulation

Abdoulaye Diamoutene, Diakarya Barro, Serge Manituo Aymar Somda, Farid Noureddine, Bernard Kamsu-Foguem. Survival analysis in living and engineering sciences. JP Journal of Biostatistics, 2016, 13 (2), pp.223-238. ⟨10.17654/BO013020223⟩. ⟨hal-02135759⟩