Využití moderních statistických metod pro analýzu nežádoucích účinků spojených s radioterapií karcinomu prostaty.

Abstract

PhD thesis on use of modern statistical methods for analysis of side effects associated with radiotherapy for prostate cancer shows, and in the particular case solves, deficiencies that arise during the collection of medical data, and focuses on ways to estimate the uncertainties in determining the incidence of observed events and the correlation analysis focused on specific biomedical data. Dissertation consists of 8 chapters. The introductory part (Chapters 2-4) is devoted to the basic theoretical knowledge about prostate cancer and its treatment, which is then going out to design software for creating and collecting medical records of patients on the course of hormone treatment and radiotherapy course and consequences (Chapter 5). Chapter 6 presents and compares different types of confidence interval for parameter of binomial distribution with respect to the analysis of biomedical data. Computing freeware applet allowing calculation of the analyzed confidence intervals is included in the attached CD. Chapter 7 is devoted to correlation analysis, especially biserial, polyserial and polychoric correlations. Parts of the attached CD are computer applets for easy calculation of the correlation between two continuous variables, two ordinal variables and between continuous variable and ordinal variable resulting discretization hidden continuous variable. Chapter 8 is devoted to application issues dealt with in Chapters 6 and 7 on the specific cancer data.