Sleep recordings: Oura vs Fitbit
|Related tools||Oura Ring, Fitbit, Open Humans|
|Related topics||Dates and Times, Sleep tracking|
Builds on project(s)
In this project I investigate how do the Oura Ring and Fitbit handle the storage of sleep data? I tried to find out by comparing data collected by a Fitbit Versa and the Oura Ring and learned that they associate sleep to different dates.
What did you do? How did you do it? What did you learn?[edit | edit source]
What did you do?[edit | edit source]
I compared sleep data for nights in 2018 and 2019 during which data was collected simultaneously by an Oura Ring and a Fitbit Versa. Initially designed to see how correlated the total amount of sleep recorded is between both of them.
How did you do it?[edit | edit source]
- wearing both devices to sleep for a bit over a year
- import data from both devices into Open Humans
- load both data into a Jupyter Notebook
- create a scatterplot for both total sleep amounts with a correlational fit
- try to improve the correlation by shifting the Oura data one day forward
The notebook for the analysis is available
What did you learn?[edit | edit source]
Doing the initial, uncorrected, correlation between the total sleep amounts of Oura & Fitbit, the results appear rather depressing, with the best linear fit being a nearly horizontal line, meaning that there is virtually no correlation between the sleep recorded by Fitbit & Oura.
Luckily, User:Madprime suggested that this might a problem of being off-by-one, which could be the result of Oura & Fitbit treating sleep differently. To test this changed the dates for the Oura ring, moving all summary dates forward from X to X+1. Doing this the correlational data immediately improves to a rather reasonable fit:
The main takeway from this: The Oura Ring stores data in a way that associates it to the day on which you went to sleep. Confusingly this even happens if you go to sleep after midnight. In contrast, the Fitbit stores the data in a way that associates it to the day on which you wake up.
This highlights the importance of understanding the Dates and Times in which your personal science data is stored, as otherwise you will be unable to make any inferences.