Hi guys,
I found several articles online and also a few threads of this forum talking about the access to fitbit data. However, all of them were out-dated, as far as I can see.
As a data scientist, I would like to export as many data from my Fitbit Surge device as possible; i.e. accelerometer data in particular. Is it possible and how? Please, only give me feedback on procedures that really worked until very recently.
Best regards,
SGAK
What do you mean by out-dated? Other than switching to OAuth2, adding a few new endpoints, and giving everyone access to their own “intraday” data without having to ask, Fitbit’s API hasn’t changed over the last few years, and the documentation is up to date.
Thanks, that’s all fine and good. But it doesn’t provide access to the acceloremeter, does it?
I don’t think Fitbit devices ever store or transmit raw accelerometer data, even internally.
Maybe you are right. I thought they do cause they are equipped and advertised with these sensors. However, maybe they only process a very short time window into the past and then forget the data instead of keeping a record. That would make sense. These are bad news for me, but thanks for this enlightening thought!
Hi Sebastian - If you want relatively unprocessed accelerometer data, why not use a DIY solution, maybe starting here:
Acceleration sensor development tools from Adafruit
I assume you want this for research, yes?
+1 to this idea.
I did exactly that using a feather board with a tiny Li-ion battery and a microSD card as storage.
I slept with the device strapped on me for many nights and got very precise accelerometer data.
Putting the collection time at 1s and very frequent data transfer (every few sec at most I think) the device ran for about 11h.
This is a good idea. Yes, you are right. It is meant for research. Thanks for the hint!
If you are going to do DIY, then I suggest that you get a full 9 DOF sensor that also gives you rotation and movement. I have started playing around with a sensor from Spark Fun 9 Degrees of Freedom - Razor IMU. It does require more hardware to load software and get the data off the board. My current setup is hardwired to a USB port on my computer.
Previously I used a 3-axis acceleration sensor (X16-1C from Gold Coast Data Concepts). It has onboard storage and is battery powered. The 12 samples per second minimum generates a lot of data over hours of use (eg sleep position).
Thank, this is a very interesting setup.
Could you share your design in more details? For example which motherboard and code did you use? Or would you have a web page where you posted those details?
The GCDC 3-axis accelerometer is used like a flash drive. Plug it in to a USB port and edit the configuration file for time and sample rate. Then unmount and collect data. The resulting files can be opened in Excel or other applications.
I am not as far along on the 9-DOF. The board is self contained you just need power and a way to get the data out of it. They provide a demonstration program you can load into the device and have a companion app to view the data.
Have you seen this thread activity-tracker-with-an-open-bluetooth it looks like they have done most of the work in the linked examples.