Semi-parametric Cox proportional hazard model and shrinkage methods

Νέστορ, Κρίστιαν

dc.contributor.author	Νέστορ, Κρίστιαν	el
dc.date.accessioned	2020-10-23T07:12:30Z
dc.date.available	2020-10-23T07:12:30Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/51606
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.19304
dc.rights	Default License
dc.subject	Ανάλυση επιβίωσης	el
dc.subject	Μοντέλο αναλογικής διακινδύνευσης του Cox	el
dc.subject	Μέθοδοι συρρίκνωσης παραμέτρων	el
dc.subject	Αποκομμένες παρατηρήσεις	el
dc.subject	Προβλεπτική ικανότητα	el
dc.subject	Survival analysis	en
dc.subject	Cox proportional hazard model	en
dc.subject	Shrinkage methods in Cox model	en
dc.subject	Censored data	en
dc.subject	Diagnostics in Cox model	en
dc.title	Semi-parametric Cox proportional hazard model and shrinkage methods	en
dc.title	Ημι-παραμετρικό μοντέλο αναλογικής διακινδύνευσης του Cox και μέθοδοι συρρίκνωσης παραμέτρων	el
heal.type	bachelorThesis
heal.classification	Survival Analysis	en
heal.language	el
heal.access	campus
heal.recordProvider	ntua	el
heal.publicationDate	2020-10-01
heal.abstract	The present thesis, which deals with the statistical analysis of lifetime data, consists of two parts. In the first part, the basic theory describing the semi-parametric Cox proportional hazard model is developed, as well as some shrinkage methods for its estimation. In the second part, this theory is applied to a specific data set. In the first chapter, some characteristics of lifetime data are mentioned. In addition, censoring, which is one of the main features, is analyzed, as well as the types that can be found in a lot of studies. In the second chapter, the basic concepts of Survival Analysis are developed, such as survival, cumulative hazard and hazard function. Also, the relationships through which these functions are connected to each other, are listed. The third chapter, concerns the non-parametric analysis of lifetime data. In other words, the procedure of estimating the survival, cumulative hazard and hazard function, without knowing the distribution of the data set, is presented. Furthermore, it includes the Log-rank test, through which we can check if there is a statistically significant difference between the survival functions of two or more groups formed by an explanatory variable. In the fourth chapter, an extensive description of the semi-parametric Cox proportional hazard model is given. Also, the procedure of fitting the model, the estimation of parameters and some diagnostics test are presented, through which both goodness of fit and the models’ assumptions are checked. Additionally, some statistical tests (Wald, Likelihood Ratio), criteria and methods for selecting explanatory variables are mentioned. In the fifth chapter, the terms Specificity and Sensitivity of a test are defined. Also, ROC curves, are analyzed; these constitute a measure of the predictive power for a model. In the sixth chapter, some extensions of the semi parametric Cox proportional hazard model are mentioned, such as the stratified model, time-depended covariates and interaction model. The seventh chapter, contains the basic theory of shrinkage methods (Ridge, Lasso and Elastic Net), through which the problem of multicollinearity is solved. Furthermore, the cross-validation method is presented. The thesis concludes with the eighth chapter, in which the theory developed in all the previous chapters is applied to a lifetime data set which consists of 500 patients who suffered from acute myocardial infraction.	en
heal.advisorName	Καρώνη, Χρυσηίς	el
heal.committeeMemberName	Στεφανέας, Πέτρος	el
heal.committeeMemberName	Παπανικολάου, Βασίλης	el
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Εφαρμοσμένων Μαθηματικών και Φυσικών Επιστημών	el
heal.academicPublisherID	ntua
heal.fullTextAvailability	false