Abstracts

An examination of the trend-renewal process for use in recurrent events modelling in sports and medicine

by Meng Zhao




Institution: McGill University
Department:
Year: 2017
Keywords: Mathematics and Statistics
Posted: 02/01/2018
Record ID: 2189605
Full text PDF: http://digitool.library.mcgill.ca/thesisfile145497.pdf


Abstract

The trend-renewal model for recurrent time-to-event data is seldom used outside of the reliability literature. This thesis thoroughly discusses the foundations of the trend-renewal process, emphasizing its applicability in the fields of sports injury and medicine. It proposes ways to better utilize a popular choice of parametric framework to address research questions in practical settings, in particular, an alternative to the classical Cox proportional intensities formulation of covariate effects. Simulation studies are carried out to evaluate the finite sample inference of parametric trend renewal models with unobserved heterogeneity. Finally, an application to a medical dataset is provided. Le modle renouvellement-de-tendance pour les donnes rcurrentes de temps de survie est rarement utilis en dehors de la littrature de fiabilit. Cette thse examine en profondeur les fondements du processus renouvellement-de-tendance, mettant l'accent sur son application dans les domaines des blessures sportives et de la mdecine. Elle propose des moyens pour mieux utiliser un cadre paramtrique populaire pour rpondre aux questions de recherche en milieu pratique, en particulier, une alternative la mthode classique Cox d'intensits proportionnelles. Des simulations sont effectues pour valuer l'infrence en chantillonnage fini de modles renouvellements-de-tendance paramtriques avec htrognit non observe. Enfin, une application un ensemble de donnes mdicales est fournie.Advisors/Committee Members: Ian Shrier (Internal/Cosupervisor2), Russell Steele (Internal/Supervisor).