Vyspělé statistické metody pro efektivní vyhodnocení biomedicínských dat

Loading...
Thumbnail Image

Downloads

10

Date issued

Authors

Janurová, Kateřina

Journal Title

Journal ISSN

Volume Title

Publisher

Vysoká škola báňská - Technická univerzita Ostrava

Location

ÚK/Sklad diplomových prací

Signature

201500562

Abstract

Survival analysis is a set of advanced statistical methods for the analysis of time to event data, typically consisting of observed event times and right-censoring times. This type of data may arise from various situations: the survival times of patients in medical trials, the lifetimes of machine components in industrial reliability, studies of duration of periods of unemployment in duration modeling. The possible fields of applications of those methods are therefore in medicine, engineering, economics, sociology or demography. In this thesis the methods of survival analysis have been used to evaluate medical right-censored data of 876 patients who underwent colectomy in the University Hospital of Ostrava. There are used two basic surgery techniques for the colectomy: either open or laparoscopic. The comparison of the two techniques in the context of mortality has been done in order to answer the question: is there a type of surgery technique which guarantees longer overall survival time? The next step has been the analysis of the uncertainty related to the risk of mortality. For analyses and comparison have been used following methods: 1. The nonparametric approach with the standard Kaplan-Meier and the Nelson-Aalen estimator of survival function including confidence interval for survival function. 2. The innovative nonparametric approach, represented by Nonparametric Predictive Inference, which uses lower and upper probabilities for quantifying uncertainty providing a model of predictive survival function. 3. The traditional log-rank test and the nonparametric predictive comparison of two groups of lifetime data, which have been compared to evaluate risk of mortality in the context of mentioned surgery techniques. 4. The semiparametric Cox proportional hazard model has been used for analyzing the influence of other patients characteristics. 5. The parametric Weibull model, used because of the fact that the hazard function of human life is often described as being “bathtub shaped”.

Description

Import 05/08/2014

Subject(s)

survival analysis, uncertainty, medical data, Kaplan-Meier estimator, Nelson-Aalen estimator, log-rank test, nonparametric predictive inference, Cox PH model, Weibull model

Citation