Sleep timeline June 24 changed the next day (and sleep stage values seem incorrect from then on)

[CKGNM6]

Hello, on June 24, the Sleep Timeline on my Fitbit app showed a low value for Deep Sleep. Surprisingly, on June 25, the Sleep Timeline Deep Sleep for June 24 was changed to 1h 32m (and the corresponding graph also changed). In fact, the Deep Sleep in the Timeline for June 25 was uncharacteristically high (2h 14m). I have lost confidence in the sleep data since June 24.

On June 25, I found the sleep.json file, which contains the data for June 24: It sill shows the >original< June 24 data for Deep Sleep. However, the UserSleepStages csv file shows the >new< June 24 data (1h 32m) for Deep Sleep. 

On June 27, I called the 1-800 number and was directed to change the Sleep Tracking to Sensitive; however, the UserSleepStages CSV data for Jun 28 still is different from the Sleep JSON data. It looks like the two files are using different algorithms to determine the sleep stages.

Also, as mentioned below, starting on June 24, the overal score in the sleep_score.csv file in the exported data also diverges from the Sleep score in the Fitbit app; prior to that their values are identical.

More info:

The UserSleepStages.csv file data up to June 23 matches the data in the Sleep.json file; however, starting on June 24, you see discrepancies such as:
"dateTime" : "02:46:00.000",
"level" : "deep",
"seconds" : 870 ← This is 14 minutes and 30 seconds, i.e., a Deep Sleep instance occured between 2:46:00 AM and 3:00:30 AM (compare with json below).

"dateTime" : "07:40:30.000",
"level" : "deep",
"seconds" : 510 ← This is 8 minutes and 30 seconds, i.e., Deep Sleep between 7:40:30 AM and 7:49:00 AM.

"dateTime" : "08:31:00.000",
"level" : "deep",
"seconds" : 30

for Sleep.json 
versus 

sleep_stage_type	sleep_stage_start	sleep_stage_end
DEEP	01:25:00+0000	01:25:30+0000
DEEP	01:26:00+0000	01:55:30+0000
DEEP	02:28:30+0000	03:01:30+0000
DEEP	04:14:00+0000	04:22:00+0000
DEEP	05:44:00+0000	05:48:30+0000

for UserSleepStages.csv

Using Python to show the first set of data (i.e., UserSleepStages csv file = Graph under Fitbit app), we get the graph below. Notice the sleep stages:

Now, below notice the sleep stages for the second set of data (i.e., the Sleep.json file):

Notice, for instance, that time intervals for Deep Sleep are different! It appears that your monthly sleep profile uses the data in the second set rather than one displayed in the Fitbit app (the first set, namely, the first graph above).

Please help!

[CKGNM6]

@ManuFitbit @MarreFitbit 

[dvdj]

I’m also having problems with sleep analysis and have noticed huge inconsistencies between the main “today” summary page and the actual sleep stage breakdown graph. I too am losing confidence in this device for sleep analysis, which is a shame because that’s the whole reason I purchased it. I hope they get to the bottom of this soon 😕

[CKGNM6]

The overal score in the sleep_score.csv file in the exported data also diverges from the Sleep score in the Fitbit app since June 24. I mean, before June 24, the values of these variables/parameteres are identical. Can anyone from the Google Fitbit software staff shed some light on this?

[danab200]

 I have had the same experience.  I have lost the confidence in Fitbit to track sleep accurately.  Keep us posted if you find another device that is dependable.

[danab200]

I have had the same experience.  I have lost the confidence in Fitbit to track sleep accurately.  Keep us posted if you find another device that is dependable.

[CKGNM6]

I would like developers to address this issue ... Why are there two different sets of data for Deep Sleep for each day?

[syj910]

My total recorded sleep decreased by 2 hours or so after June 24. Have tried all the tricks to get back to “normal”, nothing works. I have the Luxe, and everything else seems ok. Has anyone gotten answers from the company about a change to algorithm?

[CKGNM6]

It's as if the premise of this forum is for users to somehow resolve the problems abracadabra instead of developers addressing them.

What I am doing is exporting my data every day and looking through the sleep<data>.json file under the folder Global Export Data . You should get data like this (which is different from the graph in the Fitbit application):
"dateOfSleep" : 
"startTime" : "
"endTime" : "
"duration" : ,
"minutesToFallAsleep" : 
"minutesAsleep" : ,
"minutesAwake" : 

minutesAfterWakeup" : 
"timeInBed" : 
"efficiency" :
"type" : 
"infoCode" : 
"logType" : "
"levels" : {
"summary" : {
"deep" : {
"count" : 
"minutes" : 
"thirtyDayAvgMinutes" : 
},
"wake" : {
"count" : 
"minutes" : 
"thirtyDayAvgMinutes" :
},
"light" : {
"count" : 
"minutes" : 
"thirtyDayAvgMinutes" : 
},
"rem" : {
"count" : 
"minutes" : 
"thirtyDayAvgMinutes" : 
}

[whollis]

I’ve had 2 tickets opened.  They had no disclosure of such a massive change.  The other night I had a 3 hour block of time “awake” = impossible. Anybody would know if they were awake - the historical 5-10 minutes I had no issue with.  Now every day I get a poor sleep score no matter what I do.  Originally I was told it was a known app issue and they were working to fix it.

Sad since Google took over it’s a disaster.  If only the Oura ring weren’t so expensive.

[whollis]

From a helpful level 2 tech supervisor:

"If you previously tracked sleep with Fitbit, your log showed the time spent awake, restless, and asleep. We fine-tuned our mechanism for tracking sleep to estimate your sleep stages by incorporating heart rate and other data. As a result, we combined your awake and restless time into total awake minutes to give you a clearer picture of your sleep cycles.You may notice more awake minutes in your sleep stages than in your previous sleep data due to this change."

So you’ll see less sleep time and likely worsening scores due to the flawed algorithm. Word is getting out about the change and users are angry both with the new absurd methodology and having kept it a secret.  Demand a rollback like I have and others as well.

[Kris227]

I wonder if part of the problem here is timing, and rushing things...  I feel it needs time to think, and time to determine that you have actually stopped sleeping. If you check right after you get up, you will get one answer, and if you wait an hour or 2 and then check, you will get a different (I feel more correct) answer. Hope this helps. Remember that any answer is just a rough guess/approximation. Not something to live your life by!!! 🙂 A multi thousand dollar sleep lab would likely give you a different answer/opinion, but it, too, it still subject to interpreting your toss and turns during the period.

[syj910]

Thank you for the information. Needs to be fixed!!!!!
Sent from my iPhone

[whollis]

Sleep studies will yield a different answer.  However, they do not combine restlessness with awake measures which Fitbit’s new flawed algorithm does.  By relying on consistency I’ve made tweaks to my sleep habits to ensure adequate sleep.  This new scoring renders that impossible. I’ve also never had a different result 2 hours later unless somehow I edit sleep times.

When you start seeing a 2 hour “awake” phase because they’ve combined awake with restless and reduces your sleep time - do you then just ignore it?

I’ve endure some awful firmware updates, buggy recordings on the apps, etc. but this undisclosed change is really idiotic.  If they emailed us about the change and provided scientific 3rd party validation I suspect we’d feel differently about the change.  They haven’t.

[CKGNM6]

Whollis, I am the creator of this thread. How do you open tickets with Fitbit / Google? My issue in short is that there are two sets of data for the sleep stages:  One set, shown under sleep in the Fitbit app, and which reflects the data in the UserSleepStage.csv file. On the other hand, there is a second set in the sleep.json file. Now, it appear that the monthly sleep profile (The Bear, The Dolphin, etc.) is based on the sleep.json file. For instance, the Deep Sleep average for the month is basically the average of the Deep Sleep in the sleep.json file; same goes for REM sleep (if you multiply your monthly average total time in bed, I mean including awake time, by the average percentage shown, you get the average REM sleep in the sleep.json file). In fact, the percentages shown in the first set (i.e., under sleep in the Fitbit app are in fact high since June 24). I explained in my first message that the two sets appear to have started deviating on June 24.
The second set also has issues (towards the end of my sleep, if I stay put in bed, not moving for a while, the second set will mark it as Deep Sleep; the first set will not). Other than that, the second set appears to be more rigorous to show the sleep stages.
As to the Oura ring, my understading is that it is not going to be better (when bugs in Fitbit are resolved) because the Oura ring uses the same techniques to calculate the stages of sleep (movement, variability of heart rate, etc.) instead of electrodes connected to your scalp.

[CKGNM6]

I opened this thread because of the two sets of data with different values for Deep Sleep and REM. The two sets started differing on June 24. See my other posts. Because of that, one set, which contains the data for Deep Sleep shown in the Fitbit app no longer looks accurate. The other set, which is used for your Sleep Profile (i.e. an animal representing your sleep patterns) and the monthly averages therein, appears to be based on the second set. The Deep Sleep for the second set appears to be correct (or stricter than the first set) except at the end of my sleep if I just stay awake totally static (don't move) ... it gets recorded as Deep Sleep.

[CKGNM6]

I tend to focus more on the time amount of Deep Sleep and REM each night (rather than total amount sleep, e.g., time in bed minus awake/restless). I do look at the restless percentage though. In my case, if it is 6 - 8%, it is fine; if it is 10% and above, then it is indicative of something.

[whollis]

Since the changed to the new methodology it highly inflates "awake" time by combining it with the former "Restless" and as a result, your overall sleep time drops significantly.  I've had nights with 2, 3 and even 4 hour "blocks" of awake time, and the app syncs to tell me 53 POOR sleep, and awake for 4 hours.  It's just flat wrong.  I doubt Oura is using the newest Fitbit flawed methodology.  It also doesn't align with the historical HRV values - when I was 9-11 I knew it would be a bad sleep score. But now, if it's 14 - 18, I still get dinged with too many hours "awake" and a low score.

[whollis]

I would call tech support, explain your concerns over the change and ask to speak with a supervisor.  My calls have all been answered in Colombia with mixed results on being understood.  The last rep was fantastic and Cristian, the supervisor there was terrific.  He followed up, and eventually got the written acknowledgement of the change they kept hidden and shared with me.  I also spoke with a supervisior on Friday who was based in Bulgaria but was French - very easy to understand, apologetic and acknowledged that the increase in angry customers continues to grow about the flawed change.

I would start that way to get your more specific issue addressed. Do you think it is related to their June 24th change in algorithm?

[CKGNM6]

Whollis, I thought there would be a better chance to resolve your issue via another thread/topic because yours is about the lumping of "Awake" and "Restless", while this thread/topic is about two sets of data for sleep stages.