A1170
Title: Effect of sample size on precision estimates in medical studies
Authors: Svitlana Shvydka - Slovak University of Technology in Bratislava (Slovakia) [presenting]
Maria Zdimalova - Slovak University of Technology in Bratislava (Slovakia)
Abstract: The challenge of determining a reasonable sample size in medical studies is addressed. The central aim is to consider the simulation approach for evaluation of the confidence intervals (CIs) and precision (D) in samples based on empirical data (standard heart rate variability (HRV) indices (pNN50, RMSSD, PSS) and Diabetes Dataset). Two bootstrap techniques are used: the bootstrap method, changing the group membership, and the bag of little bootstraps. For the mean values, the precision is used to define the half-width of the CI as a fixed proportion of the mean (CI/2=D mean). Precision increases with sample size. However, the relative gain in precision decreases as the sample size increases. One of the reasonable approaches might be to recommend not adding further elements to the sample if the increase in precision is small. Researchers need to understand that they are trading off the precision of the estimates against the sample size of the definitive study when they decide to have an investigation with a small sample size. Svitlana Shvydka is funded by the EU NextGenerationEU through the Recovery and Resilience Plan for Slovakia under the project No. 09I03-03-V01-00029. The work is partially supported by the grant VEGA 1/0036/23.
