Séminaire scientifique - Vivien Goepp, Olivier Bouaziz
Ce séminaire aura lieu le 18 Juin 2019 de 10h00 à 11h00.
In cohort studies, heterogeneous dataset arise when the hazard rate changes with respect to the calendar time or the period time. Specific models like age-period-cohort models have been extensively studied to take into account this kind of heterogeneity. In the present talk, we present two new approaches to take into account, age, period and cohort effects. They are both based on the adaptive-ridge regularization method. The first method is applied on the SEER dataset, an American registry that comprises more than 1.2 millions of women with breast cancer, included in the study from 1973 to 2015, and followed-up from 0 to 41 years. It allows to detect heterogeneous areas for the risk of death: in particular, on the SEER dataset strong cohort effects are detected which could be the result of new public health policies that were adopted at some specific calendar times. Our second method is a direct extension of classical age-period-cohort models allowing for interactions between the three effects. The performance of this method is shown on a simulation study.