Legrand, CatherineFoulon, MathildeMathildeFoulon2025-05-142025-05-142025-05-142023https://hdl.handle.net/2078.2/32835This work studies long-term survivors with a bimodal structure in the data. Two models will be presented that allow us to take account of this particularity, which classical models fail to model adequately. These two models are the exponentiated log-sinh Cauchy cure rate and the 3-component mixture cure model. The first one has a flexible distribution capable of capturing the bimodality of the data, while the second one considers the population as composed of 3 subgroups with different risks of observing the event of interest. The different characteristics of these two models will be presented. Then, a simulation study will be conducted, to evaluate the quality of the estimates of the different parameters of these models in order to compare their performance. This work will end with an application on real data, in order to evaluate the fitting capacity of these two models and compare them.survivalcure modelsmixturebimodaltext::thesis::master thesisthesis:40892