Skip to Main content Skip to Navigation

Mesures d'influence individuelle pour modèles de régression en épidémiologie clinique

Abstract : In analyzing data collected in clinical research, the statistical models usually incorporate information provided by all the observations. The estimates, however, can rely on a small (possibly one) number of observations, illustrating their different influence. Individual influence measures have been proposed that also allow an improved understanding of the models to which they apply. The purpose of this work was to evaluate the individual influence in the setting of recent statistical models. The first section provides a measure of local influence in the proportional hazards model for the subdistribution function proposed by Fine and Gray to handle right censored data in the presence of competition. The second section of the work aims at highlighting the influence of the first individuals included in a dose finding trial (Phase I or II), using the Continual Reassessment Method (CRM). An adaptation of the CRM reducing the influence of first individuals is finally proposed.
Document type :
Complete list of metadata

Cited literature [174 references]  Display  Hide  Download
Contributor : Theses Bupmc Connect in order to contact the contributor
Submitted on : Friday, April 12, 2013 - 2:05:48 PM
Last modification on : Sunday, June 26, 2022 - 5:28:22 AM
Long-term archiving on: : Saturday, July 13, 2013 - 4:04:55 AM


  • HAL Id : tel-00812559, version 1


Matthieu Resche-Rigon. Mesures d'influence individuelle pour modèles de régression en épidémiologie clinique. Ingénierie biomédicale. Université Pierre et Marie Curie - Paris VI, 2008. Français. ⟨NNT : 2008PA066234⟩. ⟨tel-00812559⟩



Record views


Files downloads