Unexpected NW Flagging at End of Day in ActiPass

[loris-lee]

Hi,

In some datasets, ActiPass trims the end of certain days as non-wear (NW), even though those days contain wear time (e.g., ~18 hours). It’s unclear why the entire day is being classified as NW, rather than simply marked as invalid due to insufficient wear.

A screenshot from ActiPass showing trimmed days marked as NW:

A screenshot from PALanalysis v8.11.8.75 showing the same period:

The ActiPASS_BatchSettings.txt file from the relevant batch:
ActiPASS_BatchSettings.txt

Questions:

1. Could you clarify why ActiPass classifies an entire day as NW in such cases?
2. Is there a way to retain the NW classification for only the actual non-wear periods?

Thanks you!

[pasan-h]

Hi,
Good observation :-).
In this case, ActiPASS auto trimmer has trimmed the second wear segment because it is less than 48h. ActiPASS auto trimmer trims wear segments which are less than 48h by default (unless it is the first segment). Due to various reasons ActiPASS default behaviour is that it expects long wear segments. As can be seen from PAL analysis output the participant has a non-wear segment (on Saturday) followed by a relatively short wear segment. What has happend is the expected behaviour under this situation. However in the future the minimum duration for a wear segment might be changed (with a new release).

However we can disable the auto trimmer by explicitely defining measurement start/stop times by using diary with predefined "Start" and "Stop" events. See ActiPASS diary documentation for more info. When using a diary make sure the IDs specified are matched with IDs ActiPASS detects from file names. Therefore I suggest changing the ID format to "Start" and 7 digits (IDs in format PL#####). I think now you have set ID format to "full filename". See ActiPASS userguide.

We should be be aware of another possibility. Due to undocumented changes to PAL DATX file format, the open-source DATX file parser might fail to fully read DATX files (not in this situation though I think). In rare occations I have heard about partial file loading issues. SInce PAL has decided to not to open their binary file format unlike most other manufacturers only way we can completely avoid file-loading issues is that we first convert all files to CSV. I know CSV files are huge 😊, but that is the penalty researchers pay for adopting proprietery systems.

Also please keep in mind that ActiPASS and PAL analysis might differ in detecting non-wear (as well as other activities). NW according to one system may not be flagged as NW by the other under certain situations. So don't accept the output of PAL software as ground truth. NW detection is another situation accelerometers with temperature sensors (like Axivity) can do better.

All the best with your anlysis!
/Pasan