Analysis of Accelerometer Data With Nonvalid Days Using Pattern-Mixture Mixed-Effects Models
Journal for the Measurement of Physical Behaviour · 01.01.2026 · 9 (1)
DOI: 10.1123/jmpb.2025-0079 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal for the Measurement of Physical Behaviour, erschienen am 01.01.2026. Verfasst von Xingruo Zhang und 3 weiteren Autor:innen. Die Arbeit nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
In the analysis of accelerometer data, a common practice is to exclude days when participants wore their accelerometers for less than some threshold (e.g., 10 hr), also known as nonvalid days. This practice, however, introduces the potential for selection bias and diminished statistical power since only partial data are considered. This article introduces pattern-mixture mixed-effects models as a robust analytical tool tailored to accelerometer data with nonvalid days, with adjustments made to accommodate the time-varying pattern of missingness. We apply pattern-mixture mixed-effects models to accelerometer data from a randomized controlled trial of lifestyle intervention. Simulation studies are conducted to validate these models and further illustrate their benefits. The use of pattern-mixture mixed-effects models for analyzing accelerometer data with nonvalid days yields more precise parameter estimates with reduced biases as compared with standard mixed-effects models using only valid days or all observations. Our findings underscore the potential of pattern-mixture mixed-effects models as a robust and precise method to analyze accelerometer data.
Zitieren
APA
Zhang, X., Siddique, J., Spring, B., & Hedeker, D. (2026). Analysis of Accelerometer Data With Nonvalid Days Using Pattern-Mixture Mixed-Effects Models. Journal for the Measurement of Physical Behaviour, 9(1). https://doi.org/10.1123/jmpb.2025-0079
Vancouver
Zhang X, Siddique J, Spring B, Hedeker D. Analysis of Accelerometer Data With Nonvalid Days Using Pattern-Mixture Mixed-Effects Models. Journal for the Measurement of Physical Behaviour. 2026;9(1). doi:10.1123/jmpb.2025-0079
BibTeX
@article{Zhang2026,
author = {Zhang, Xingruo and Siddique, Juned and Spring, Bonnie and Hedeker, Donald},
title = {Analysis of Accelerometer Data With Nonvalid Days Using Pattern-Mixture Mixed-Effects Models},
journal = {Journal for the Measurement of Physical Behaviour},
year = {2026},
volume = {9},
number = {1},
doi = {10.1123/jmpb.2025-0079}
}
EUVIMED