I'm using a software developed for smartphones to estimate the time spent in different physical activity intensities based on cut-points established for triaxial accelerometers. But even with previously developed studies that suggest a relative equivalence of "counts" measured by Smartphones and Actigraph accelerometers (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4414958/), I'm not sure if even with an equivalent sampling rate I can use cut-points for "counts" established for triaxial accelerometers. Can I do that, or are the "counts" estimated through the smartphone different from those estimated by actigraph accelerometers, even for an equal sampling rate?
Hi Douglas, according to my experience, sampling rate is not the only one factor that affect the 'count' or cut point establishment. Other than that, the placement of the sensor and the age of the population are also factors that can change the cut-point threshold. In addition, the way you preprocess the accelerometer data is also affect the result, for example: the paper threshold was based on filtered data but your accelerometer data was not filtered then the result may not valid.
There are many papers develop their own cut-points based on different sensors, placement, and target population. You can check below reference for more detail:
https://github.com/digitalinteraction/openmovement/wiki/AX3-GUI
Article Large Scale Population Assessment of Physical Activity Using...
- Badges
- Science topic