.. _preproc-non-wear-detection-top:

===================================================
Automatic Detection of Non-Wear Segments Using GGIR
===================================================

Cicada has implemented an algorithm adapted from the GGIR package to automatically detect non-wear segments.

.. note::

    **Reference:** Van Hees, Vincent T., et al. "Separating movement and gravity components in an acceleration signal and implications for the assessment of human daily physical activity." PloS one 8.4 (2013): e61691.

The algorithm uses a 60 minute window that slides across the data in steps of 15 minutes. For each iteration, a window is marked as 'non-wear' if the standard deviation AND the range of the raw acceleration data is less than their respective thresholds for at least two out of the three directions ('x', 'y', and 'z').

**To automatically detect non-wear segments,**

- click ``Preprocess`` > ``GGIR Non-Wear Detection``.