Automated splitting into batches for observational biomedical studies with sequential processing
Authors:
Open Access
- 10 May 2022
- journal article
- research article
- Published by Oxford University Press (OUP) in Biostatistics
- Vol. 24 (4) , 1031-1044
Abstract
Summary: Experimental design usually focuses on the setting where treatments and/or other aspects of interest can be manipulated. However, in observational biomedical studies with sequential processing, the set of available samples is often fixed, and the problem is thus rather the ordering and allocation of samples to batches such that comparisons between different treatments can be made with similar precision. In certain situations, this allocation can be done by hand, but this rapidly becomes impractical with more challenging cohort setups. Here, we present a fast and intuitive algorithm to generate balanced allocations of samples to batches for any single-variable model where the treatment variable is nominal. This greatly simplifies the grouping of samples into batches, makes the process reproducible, and provides a marked improvement over completely random allocations. The general challenges of allocation and why good solutions can be hard to find are also discussed, as well as potential extensions to multivariable settings.Keywords
Funding Information
- The Bergen Research Foundation
- Helse Vest’s Open Research Grant (#F-12144)
- Research Council of Norway (#301178)
This publication has 15 references indexed in Scilit:
- Transformation, Normalization, and Batch Effect in the Analysis of Mass Spectrometry Data for Omics Studies2016
- Randomization in Clinical Trials2015
- UK Biobank: An Open Access Resource for Identifying the Causes of a Wide Range of Complex Diseases of Middle and Old AgePLOS Medicine, 2015
- Cohort Profile: The HUNT Study, NorwayInternational Journal of Epidemiology, 2012
- OSAT: a tool for sample-to-batch allocations in genomics experimentsBMC Genomics, 2012
- Design of Comparative Experiments2008
- The biobank of the Norwegian mother and child cohort Study: A resource for the next 100 yearsEuropean Journal of Epidemiology, 2006
- Association Schemes2004
- The Optimal Design of Blocked and Split-Plot Experiments2002
- Reducibility among Combinatorial Problems1972