Background

Accelerometers have become a popular assessment tool of physical activity over the last decades. The small body-worn sensors provide an easy and more objective alternative to classic questionaire-based assessment while simultaneously keeping the researcher and participant burden low. Especially, the field of raw data accelerometry, the analysis of the raw acceleration signals measured in g (1 g = 9.806 65 m s−2), has received great focus over the last years and is a rapidly advancing field. Many algorithms have been proposed the last years; Also by our lab. Simultaneously, openly available data to benchmark algorithms on is scarce due to privacy concerns. Nevertheless, new algorithms can only be adopted in research after rigorous external validation.

While publishing code and, in the context of machine learning, trained models has become more common, this often does not automatically imply that the published code is easily usable for validation. Effectively, often reimplementations are necessary, even though they increase potential biases by incorrect implementation. For that reason, we developed paat as a simple and easy to use package to facilitate replicating and validating of our findings and prospectively to apply the algorithms in research. The package is structured according to the respective applications (io, preprocessing, features, wear time, sleep, estimation) and the methods easily applicable also in isolation. An overview over the different submodules can be found in the API Documentation.

However, paat has already been used in various studies. Syed et al. [1], for instance, developed and used the general gt3x reading functionality and implemented and used the NWT algorithm from Van Hees et al. [2] for a comparison study of different NWT algorithms. Syed et al. [3] also used the functions to develop a new non-wear time algorithm which is now included in paat. Weitz et al. [4] used the package to load and process the acceleration data to investigate the effect of accelerometer calibration on physical activity in general and MVPA in particular. Weitz et al. [5] used the package to load and process the data in order to train a machine learning model to identify time-in-bed episodes.

If you are using paat in research, feel free to cite it as

Weitz, M., Syed, S., & Horsch, A. (2024). PAAT: Physical Activity Analysis Toolbox. Zenodo. https://doi.org/10.5281/zenodo.13885749

If you are using BibTex you may want to use this example BibTex entry:

@misc{weitz_paat_2024,
        title = {{PAAT}: {Physical} {Activity} {Analysis} {Toolbox}},
        url = {https://doi.org/10.5281/zenodo.13885749},
        publisher = {Zenodo},
        author = {Weitz, Marc and Syed, Shaheen and Horsch, Alexander},
        year = {2024},
        doi = {10.5281/zenodo.13885749},
}

This also helps us to keep this page up to date.

---

[1]

Syed, S., Morseth, B., Hopstock, L. A., & Horsch, A. (2020). Evaluating the performance of raw and epoch non-wear algorithms using multiple accelerometers and electrocardiogram recordings. Scientific Reports, 10(1), 1–18. https://doi.org/10.1038/s41598-020-62821-2

[2]

Van Hees VT, Renström F, Wright A, Gradmark A, Catt M, et al. (2011) Estimation of Daily Energy Expenditure in Pregnant and Non-Pregnant Women Using a Wrist-Worn Tri-Axial Accelerometer. PLOS ONE, 6(7): e22922. https://doi.org/10.1371/journal.pone.0022922

[3]

Syed, S., Morseth, B., Hopstock, L. A., & Horsch, A. (2021). A novel algorithm to detect non-wear time from raw accelerometer data using deep convolutional neural networks. Scientific Reports, 11(1), 8832. https://doi.org/10.1038/s41598-021-87757-z

[4]

Weitz, M., Morseth, B., Hopstock, L. A., & Horsch, A. (2024). Influence of Accelerometer Calibration on the Estimation of Objectively Measured Physical Activity: The Tromsø Study. Journal for the Measurement of Physical Behaviour, 7(1). https://doi.org/10.1123/jmpb.2023-0019

[5]

Weitz, M., Syed, S., Hopstock, L. A., Morseth, B., Henriksen, A., & Horsch, A. (2025). Automatic time in bed detection from hip-worn accelerometers for large epidemiological studies: The Tromsø Study. PLOS ONE, 20(5), e0321558. https://doi.org/10.1371/journal.pone.0321558