Ανάλυση του χρόνου επιβίωσης ασθενών με καρκίνο του μαστού μετά από χειρουργική επέμβαση
Μάρκο, Ερνίντα Μ.
The purpose of this study is to analyze the data used in medical statistics, in order to study the survival time in women with breast cancer. The analysis of survival time of patients will rely primarily on Cox’s proportional hazards model. But to get a complete picture of overall survival analysis for these data, first we define the basic concepts on which survival analysis is based on, while reference is made to Kaplan - Meier estimator and to the basic survival functions test. Then we consider both theoretically and practically Cox’s proportional hazards model. Variable selection techniques are described and analyzed in order to identify the best model for the interpretation of our data. Subsequently we check if the model is adequate, if it satisfies the proportional hazards assumption. This assumption is controlled by graphical methods, Schoenfeld and Scaled Schoenfeld residuals and through other statistical tests. Finally, aiming at a better model both in terms of adaptation and proportionality we define and implement frailty models. The data analysis was conducted with the implementation of the program R.