A.V. Ivanova1, P.E. Soldatov2, Yu. I. Voronkov3, A.M. Nosovsky4

1–4 Federal State Budget Institution of Science State Scientific Center of the Russian Federation –
Institute of Medical and Biological Problems of the RAS (Moscow, Russia)
1 iivalexa@yandex.ru, 2 marquet@yandex.ru, 4 collega1952@yandex.ru

In today’s world, scientific research plays a key role in the development of medicine and other fields of science. However, for research results to be credible and accepted by practitioners, a rigorous statistical underpinning of the findings is required. The principles of evidence-based medicine set high standards for the reliability of comparative evaluation of research results.

In the context of medical research, statistical methods are particularly relevant for a small number of observations, which may be common to rare diseases or unique medical cases. Statistics provide tools to analyse even small samples, allowing conclusions to be drawn with a reasonable degree of confidence.

Based on developed statistical methods, such as the t-distribution of Y. Gosset and the F-distribution of R.A. Fisher, as well as non-parametric methods, it is possible to analyze the results of studies even with a small sample size. Starting with df=30, the close approximation of t-distribution to the normal distribution makes it possible to use statistical methods even in small samples.

However, it is important to keep in mind possible errors that may occur in statistical data processing. Registration errors and representativity errors can distort research results. Random sampling errors decrease with increasing sample size, while systematic errors such as bias error may increase with increasing sample size.

An important aspect of statistical analysis is the assessment of the level of relevance and strength of statistical criteria. The significance level allows you to estimate the probability of a Type I error, and the strength of the criterion is the probability of detecting differences when they actually exist. The correct choice of a statistical criterion and the estimation of its capacity are important for the credibility of the study findings.

The use of different statistical criteria, such as shear and scatter, allows for the analysis of differences between samples or groups. For example, the Wilcoxon criterion and its analogs can be used to compare medians or to estimate differences in sample shear.

In conclusion, statistical data processing plays a key role in scientific research, especially in medicine. Rigorous statistical substantiation of research results allows drawing conclusions that can be adopted by doctors and specialists. Assessing the level of relevance, the strength of the criteria, and selecting a suitable statistical method are important steps in planning and conducting research.

Ivanova A.V., Soldatov P.E., Voronkov Yu.I., Nosovsky A.M. Statistics of small samples in biomedical studies. Biomedicine Radioengineering. 2025. V. 28. № 7. P. 26–32. DOI: https:// doi.org/10.18127/j15604136-202507-03 (In Russian)

1. Bol'shev L.N., Smirnov N.V. Tablicy matematicheskoj statistiki. M.: Nauka. 1995. 416 s. (In Russian).
2. Korn G., Korn T. Spravochnik po matematike dlya nauchnyh rabotnikov i inzhenerov. M.: Nauka. 2003. 832 s. (In Russian).
3. Kobzar' A.I. Prikladnaya matematicheskaya statistika. Dlya inzhenerov i nauchnyh rabotnikov. M.: Fizmatlit. 2006. 816 s. (In Russian).
4. Hollender M., Vul'f D.A. Neparametricheskie metody statistiki. M.: Finansy i statistika. 1983. 456 s. (In Russian).
5. Nosovskij A.M. Primenenie veroyatnostnyh modelej na okruzhnosti v mediko-biologicheskih issledovaniyah. Kosmicheskaya biologiya i aviakosmicheskaya medicina. Tezisy dokladov IX Vsesoyuznaya konferenciya Kaluga (19–21 iyunya 1990). 1990. C. 147 (In Russian).
6. Praveckij N.V., Nosovskij A.M., Matrosova M.A., Holin S.F., SHakin V.V. Matematicheskoe obosnovanie dostatochnogo kolichestva izmerenij dlya dostovernoj ocenki registriruemyh parametrov v kosmicheskoj biologii i medicine. Kosmicheskaya biologiya i aviakosmicheskaya medicina. M.: Medicina. 1990. № 5. S. 53–56 (In Russian).
7. Nosovskij A.M., Praveckij N.V., Holin S.F. Matematicheskij podhod k ocenke tochnosti izmerenij fiziologicheskogo parametra razlichnymi metodami. Kosmicheskaya biologiya i aviakosmicheska medicina. M.: Medicina. 1991. № 6. S. 53–55 (In Russian).
8. Nosovskij A.M., Savina N.V., Osipov Yu.Yu. Filipenkov S.N. Primenenie metoda singulyarnogo spektral'nogo analiza vremennyh ryadov s propuskami dlya ocenki adaptacionnyh vozmozhnostej organizma pri vnekorabel'noj deyatel'nosti kosmonavtov. Biomedicinskaya radioelektronika. 2009. № 3. S. 9–13 (In Russian).
9. Bland J. M., Altman D.G. Measuring agreement in method comparison studies. Statistical Methods in Medical Research. 1999. V. 8. P. 135–160.
10. Jensen A.L., Kjelgaard-Hansen M. Diagnostic test validation. Schalm's Veterinary Hematology. Ed. by D. Weiss, K. J. Wardrop. 6th ed. Ames: Wiley-Blackwell. 2010. P. 1027–1033.
11. Bland J.M., Altman D.G. Statistical method for assessing agreement between two methods of clinical measurement. The Lancet. 1986. V. 1 (8476). P. 307–310.