BAYESIAN SAMPLE SIZE DETERMINATION FOR CYTOGENETIC STUDIES IN A POPULATION

Bayesian approach for the sample size determination (SSD) and a comparison with classical (Frequentist) approach are presented. Credible interval length estimation-type criterion was applied for the Bayesian SSD estimation in population studies of cytogenetic characteristics. The dependence of the sample size (n) on the length of the 95% Credible interval of the population mean has been estimated in the Gaussian approximation of the distribution functions with known variance and an unknown population mean. The Mean and Variance of the prior function in the Bayesian approach were estimated based on published data and the results of our previous studies. Mathematical analysis and graphical visualization of the results was carried out using the software STATISTICA-12, and WinBugs. It is shown that the Bayesian approach achieves an almost two-fold decrease in sample size and provides the possibility of flexible optimization of the planned procedures at the preliminary stage of the study. Further increase inaccuracy of the results is expected due to a more accurate approximation of asymmetric distributions using gamma functions.