1. V.I. Vernadsky Institute of Geo and Analytical Chemistry of Russian Academy of Sciences 2. Moscow State Taxing Academy 3. Botkin Clinical Hospital
Type: Experimental/clinical study
Year: 2004 vol: 50 application: 1
Abstract: A relatively small sampling sets of biomedical data lead to great difficulties in their statistical processing because of assuming of normal distribution low of probabilities. Standard biometrical methods are based on such a low. But results of biomedical experiments as a rule belong to more complicated lows. It is quite serious restriction. But we may overcome it with bootstrap method. This paper describes special experiments proving bootstrap method convergence in the case of evaluation such statistical functions as mean value, standard deviation, asymmetry, access and correlation coefficients for multidimensional data of biomedical origin. It is shown that bootstrap results are in better agreement with a nature of biometrical sense than results of classic statistical methods for small data arrays. Bootstrap method may be recommended for wide using in biomedical data processing
Reference: Dementiev V.A., Soroka A.V., Khimochko T.G., Bootstrap features in retrieving of complicated statistical functions for small sampling sets in biological and medical investigations, Biomeditsinskaya khimiya, 2004, vol:
50(application 1), 117-126.