Manuálny Výpočet Deskriptívnej Štatistiky, Intervalov Spoľahlivosti, Korelačnej Analýzy, T-Testov, F-Testu, Multivariačnej Analýzy a Aanlýzy prežívania V Ére Softvéru Typu "Black Box" Pre Hodnotenie Zmeny Zdravotného stavu Pacienta

Abstrakt

Starting point: Modern statistical models are essential for methodologically sound evaluation of patient health status changes and precise prognostication. Between 2003 and 2025, the scientific community has seen a rise in incorrect statistical application. Reliance on "black-box" software, artificial intelligence, or automation without thorough understanding of key parameters often results in inaccuracies. Advanced statistical software streamlines calculations but does not replace model comprehension.

Group: We evaluated health status changes before and after treatment and rehabilitation in two distinct populations: wrist fracture patients and those with intramedullary osteosarcoma. Post-fracture muscular recovery demands precise diagnostics, particularly the standardised Janda test for reliable progress monitoring. Conversely, osteosarcoma rehabilitation prioritises verticalization training to optimise centre of gravity control and activate postural muscles. This approach maintains maximal sample heterogeneity, allowing monitoring of changes in both non-life-threatening and life-threatening diseases.

Methods: We propose a methodological approach incorporating a value matrix based on fuzzy logic at various significance levels to assess subjective, quasi-subjective, and objective health parameters. Consequently, we have detailed step-by-step manual calculations to ensure methodological clarity.

Results: Our study found that 65 out of 600 scientific papers employed inaccurate or not-optimized statistical models.

Conclusions: The aim of this work is to provide a methodological guide for the accurate application of advanced statistical models in scientific research, particularly for early-career physicians. Additionally, it aims to inform experienced professionals about potential ambiguities and procedural steps when utilising "black-box" software.