Generating hourly-level physical activity data

[Kaiyi-Jin]

Hi ActiPASS team!

I'm Kaiyi from National University of Singapore. We're exploring physical activity patterns and are particularly interested in hourly-level activity metrics (beyond daily aggregates). Could you advise if:

1. Current ActiPASS models/outputs support generating temporal activity breakdowns at hourly resolution?
2. If not natively supported, would preprocessing raw sensor data (e.g., accelerometer) with custom time windows be a viable workaround?

Any documentation references or implementation suggestions would be greatly appreciated!

Thanks again for building this open-source tool!

[pasan-h]

Hi Kaiyi,
I think this is an interesting question. You have two options. One is pretty straightforward and the other needs some more work.

ActiPASS already provide some information about hourly behaviour patterns for weekdays and weekends. These are saved inside "IndividualOut". Have a look at files ending "*** - Activity Histogram-Weekends.csv" and "**** - Activity Histogram-WeekDays.csv". If we set "Visualizations" to "full", we can also see the hourly activity visualizations as shown below.

The other option is to use the ActiPASS diary. ActiPASS supports interval-based analysis using an additional diary. The idea of an ActiPASS diary is that we can divide a measurement into intervals (which corresponds to different context like "Work", "Leisure" for an example). Since the diary format is flexible we can easily create a diary which divide the day into 24 hourly segments.

1. ActiPASS has a sequential diary format (i.e. each interval lasts until the start of next event). See more info about ActiPASS diary here: https://github.com/Ergo-Tools/ActiPASS/wiki/ActiPASS-diary-file-format

2. Create a diary like shown below. It's a good idea to use the special events "Start" and "Stop". See above.

   

3. Make sure IDs you specify in diary are matched with what ActiPASS considers as ID.

4. Set ActiPASS in to advanced-mode

5. Set output table format to "EventsNoBreak" in advanced settings. See below screenshot.

   

6. ActiPASS will then create an interval-based output table. Use this table for further analysis.

7. Since above table contains no quality-check data be very cautious using this. It's a good idea to consider the daily-quality-checks flags from the daily table first and then only consider days which are not flagged as bad.

[Kaiyi-Jin]

Thank you so much. The diary method is smart!

[Kaiyi-Jin]

Dear Pasan,

Thank you for your idea of using diary! I did a test and it went smooth with your detailed guideline.

However, I found that the output file EventAverageTimes is not overlaid with timestamps. Is there any way to generate such information? Any idea, meta data would be appreciated!

Thanks in advance!

Kaiyi

[pasan-h]

I exactly don't understand your question 😀. The variables based on diary intervals are found in ActiPASS table "ActiPASS_Events_LongFormat_MasterFile.csv". See 6 above.