Robust Methods in Statistics and Their Implementations for the Support and the Analysis of the Results of Randomized Clinical Trials

Every year new drugs, therapies, and treatment protocols are developed. Officially, they have to pass a special test for statistically significant efficacy and safety before using in real practice. To prove it, clinical trials are conducted - scientific experiments aimed at determining the level of efficacy, safety of a drug or therapy. That is why the medicine based on such rigorous statistical results called evidence-based. Evidence-based medicine can be considered as an interdisciplinary approach to the improvement of healthcare system because it requires not only highly qualified medical staff, but also experts in the field of statistics and design of experiments. As a result of their joint work in past decades, a variety of mathematical methods and models to analyze the results of such trials was created. One of the main goals of this work - to develop such application for survival analysis of a raw data from clinical trials. In addition, several tasks were stated: to analyze the current state of both medical statistics and software tools based on these methods, taking into account their historical origins and development. In particular, emphasis was placed on robust methods, which are becoming more popular. As a result, in this paper basic mathematical methods for the analysis of the results of clinical trials and additional robust algorithms for dealing with biases and possible inaccuracies in the input data were considered. The description of the applicability and working algorithms were made. Software implementation on the programming language R was created in the form of a new package to it. It allows the basic descriptive statistics of a given data set, as well as the most useful survival analysis procedures, while requiring minimal operations by the physician.