Estimated Oxygen Variation graph is broken

[AndreaVersa2can]

Hi, I just noticed that the green and yellow lines have changed and now are all over the place on the graph. All graphs for different days start with a 0:00 hrs. The fluctuations are not at all in line with what I was reading up until today.  Anyone experienced this issue?? Thanks

Moderator Edit: Clarified subject

[SunsetRunner]

The increased granularity shows the momentary spikes which previously were under the smooth curves. That seems to be the only difference. They are presenting the data every minute instead of every 5 minutes (for example, there's no way of knowing) and this shows those spikes.

Note that increased granularity does not mean it's more accurate in terms of actual oxygen variation from the larger time scale which is what matters physiologically. The increased granularity could also, and in my opinion, has, introduced a significant level of noise to the data.

Note also please that I am stumbling around in the dark too, as Fitbit refuses to label the axes. What I am speculating is based on decades of knowledge of data analysis in science.

[SigHillDude]

I'm back to having sync issues too.  Painful.  I don't knwo why they can't get the basics right.  Reboot phone, reboot device, turn bluetooth on and off.  Maybe they should actually send people to do syncs.  Might encourage them to get it fixed. 

[MarkFolse]

The lack of a scale on that graph is part of my complaint.

Given the variation in display between IOS and Android I have no idea what
my Android is telling me.

Moderator Edit: Formatting

[pforsche]

Just a thought:

It looks like a smoothing function has been applied to the data in the iPhone plot -like a rolling average. Maybe what we are all looking at on our Android phones is the raw EOV data; if so, it's pretty noisy. Some relevant information from the Android developers would be useful.

[Shivparsh]

If the least count is now 1 minute instead of 5, the graph spread would have been wider. 

In any case, Fitbit should have clarified changes post the upgrade.

[SunsetRunner]

The graph spread will not be wider unless the X scale changes. The big issue here is the loss of the smooth curve.

[HDWorker]

SO THE FIX IS?????

[LizzyFitbit]

Hi everyone, and welcome to our new members.

Thanks for the details provided, as well as your feedback. I apologize for this situation experienced with the Fitbit app. As mentioned before, our team is working on a new Fitbit app version which includes improvements for the Estimated Oxygen Variation. This version will be available this month, so please make sure to keep the Fitbit app updated as you've been doing so far. We'll also update the Android board once it's released.

As a friendly reminder, we want our Fitbit Community to be a place where users can come to share their feedback and stories in a conducive manner, so please keep in mind the Community Guidelines when posting.

[DannyFB]

Yup.

Hence my fascination with the huge kickback on the current bug with displaying whatever those discrete data points ever were. The chart has long been of extremely vague utility.

[LindainSD]

It would be nice if you would acknowledge to you customers that you accidentally broke a feature and that you are sorry,  but that never happens either...so good luck having customers who aren't pissed off when your company is completely arrogant.  I guess we should take our feedback to more public forums since you are concerned about what we post directly to you. Sent from my Verizon, Samsung Galaxy smartphone

[LindainSD]

High variation in sleep apnea is when your oxygen % changes by 4% or more.   Changes of less than 4% are normal when sleeping.   I'd like to think they would have stuck with the industry on this,  but who knows.   The point people were trying to make is suddenly it drastically changed with zero explanation.  

What I can tell you since I actually wear an oximeter at night is the previous graphs were more accurate in terms of industry terms and actually measurements.   I have no idea what they are graphing now... potentially actual oxygen,  but if so the graph is backwards (upside down).

[SunsetRunner]

But there is no way of telling what the variation is because the Y axis is unlabeled.

[TDundy]

Wow that is amazing. Wish I could do that. I bet yr chart looks totally
different now .

[4Peace54]

When can we expect a correction of this issue? Like others, the main reason I have a fitbit is to monitor my apnia. This is an important issue for me and it would be helpful to be notified when corrections have been made.

[TR51]

I'm not surprised that you made an appointment. It was the first thing that came to my mind when I noticed this issue yesterday, i.e., I wonder how many people will all of a sudden schedule an appointment with their doctor and then be referred on for an expensive sleep study.

[ItsFe]

Yes, saw this and scared me to death. Immediately thought I needed a sleep study 👀

[SigHillDude]

The chart is EOV. It is not SpO2.  It is not live data readings translated to a bar or line graph. It is a manipulation of the data that has no defined values, no time frame, etc. It is not simply a change in the time access and a switch from a line graph to a bar graph, it is a computational change as well. Some similarities may be drawn but there are differences in the chart as well.  One can not smooth the bar chart to derive the same line chart. This can only be seen if you examine the same data on the new release and an older release.  

Prior, I raised an issue with Fitbit that there are flat horizontal lines in the line graph, or no bars in the bar graph.  These are not exactly aligned in the two formats. Further, the flat line is not a zero variation as there are also lines/bars extending below the flat sections. Without further information on the algorithm from Fitbit, one can not project what anything means as the maximum SpO2 is 100% and Fitbit declares they do not report values under 80%.  Given the narrow 20% window and not getting an average SpO2 reported, what does variation mean.  Is high variation approaching 80% or is it average less 1-2%?  What does it mean when variation drops into the negative zone, that it is less than zero variation or that the algorithm is simply flawed.  And why not provide time labels that match and align with the same horizontal time labels on the sleep chart above?  Or better yet, why don't they simply report real time SpO2 as they declare it is being measured all night long anyway......

---

It is not simply the loss of the smoothing function and display. One has to directly compare the charts to see that the displayed data is different. 

---

And now even more vague.  

Moderator edit: format

[SunsetRunner]

As I said on another thread, yes, the graph is a first derivative graph of spO2, otherwise known as EOV. I have complained about lack of axis labels also. Real time spO2 would make more sense for somebody who suspects sleep apnea.

[SigHillDude]

They won't declare over high is greater than 4% and what is the negative variation?  If 4% is normal, I guess I am not normal because I can perk my spO2 from 89% to 96% with a couple breaths when I am relaxing and the nurse is monitoring my SpO2.

[Bubbabighatt]

I too am one of the many who was very startled to see how poorly I was breathing.  But hey, I can't complain because after all as soon as Fitbit discovered there was a problem with their programming they sent out that email explaining what happened and how and when they were going to fix it. 

Great customer service, Fitbit!  As always!  The forgoing was sarcasm - They didn't send out anything.