spo2 chart is wrong - Fitbit not providing explanation

[juliefouc]

This post has two parts: 

1) 

My avg spo2 last night was 95. The range was 95-96. How is the variation chart as 'eventful' as the ones I get when my spo2 ranges from 93 to 99? Straight horizontal lines suggest no variation for that time period. But how is it possible there would then be a dip after that line? It's impossible to have less than 'no variation'. Further, no variation should be at the very bottom of the y axis (so the dimension is also completely nonsensical). The parameters behind that chart need to be made transparent. 

Of note, I'm a statistician, so unless I'm missing something here, I'm not talking out of my butt here. 

2) 

How does one make a complaint to Fitbit? Their customer service is ATROCIOUS!!! I've tried asking about this through chat, but all they are able to tell me is to make sure my watch is well-adjusted on my wrist, etc. while I sleep. Which has absolutely nothing to do with the chart. I'm not debating the values themselves, but rather, the analytics of the chart itself. There doesn't seem to be an escalation process available (how is that even possible in this day and age?) and it seems the only other option is to contact Better Business Bureau, which I really would like to avoid. 

[Guy_]

Hi @juliefouc - It may be the "Estimated" Oxygen Variation and the SpO2 measurements are done at a different frequency using different measurement techniques.

The EOV is always measured, whereas the SpO2 has it's own App, or clock face, and the aggregating of the information is done differently from the different readings. The SpO2 calculation and presentation is done at Fitbit and the EOV seems to be based on information done locally, in the Fitbit App, or even possibly on the watch or both.

[juliefouc]

Thank you for the reply. Do you know, then, how the graph is actually made? (eg. what the parameters are to determine what constitutes as high vs. low variation?). And why you can get a dip lower on the 'low' axis after having had a straight horizontal line slightly higher up? That's the part that makes no sense (at least to me). The numbers themselves are not what I don't agree with it. They're not even really relevant to the present issue...It's the fact that the chart seems to have some kind of third axis they haven't defined. A straight line would mean 'no variation', correct? Not high, not low, just static. That's what straight lines on line charts usually indicate. Then how is it possible to get less than 'no variation'? It's not. And that's what Fitbit needs to explain. 

[Guy_]

@juliefouc- Not sure there is any publication on how the information is processed.

One can imagine the absence of a reading could be an issue, along with an issue in the Estimation process. The EOV information is probably only indicative at best.

Someone more experienced might be able to give you more information if you posted a photo of the anomalies you observe.

[juliefouc]

Thanks : ) 

You might be onto something... maybe the straight line is when there isn't a reading (ie. if I'm lying on top of it?). That's the only thing that would make any sense. 

I am trying to attach a photo but it doesn't seem to be working... Seems others are having the same issues trying to attach a photo from a windows browser. I'll try to do it from my phone in a subsequent 'reply'. 

[juliefouc]

Thanks : ) 

You might be onto something... maybe the straight line is when there isn't a reading (ie. if I'm lying on top of it?). That's the only thing that would make any sense. 

I am trying to attach a photo but it doesn't seem to be working... Seems others are having the same issues trying to attach a photo from a windows browser. I'll try to do it from my phone in a subsequent 'reply'. ---

[Guy_]

@juliefouc- If you need more detailed information you can always export all your data from your Web dashboard, "Request data" option, but that;'s quite a lot of information.

Look in the Sleep and Physical Activity folders for the raw data which you may be able to massage in a way that provides some form of explanation.

[juliefouc]

 

[juliefouc]

I managed to post the picture! (lol that was quite exciting considering how complicated it was!)

As you see, it's not a problem with the data itself. I don't actually care about the numbers at this point. What I care about is the fact that the values on the chart (variation... not numbers) seem to be pulled out of thin air. The way it is presented is a logical fallacy (unless the horizontal line means there was no recording at that time... that's the only thing that makes sense... and if that's what this is, then Fitbit should really provide some kind of legend/ explanation as to how to read the values... I realize the layperson might blindly accept a chart provided by such a reputable company, but I find the lack of transparency alarming). At the end of the day, I don't care about the readings. I have doctors for that kind of stuff. But Fitbit is presenting this chart as accurate and they need to at least be able to explain what the values mean/ how they were computed. This isn't a big or unreasonable ask. 

[Guy_]

@juliefouc- Well done, the photo looks like good sleep.

Bear in mind that breathing rate is used as part of the Estimation process - which can therefore be used to detect apnoea.

[juliefouc]

Thanks, yes, I've been sleeping relatively well lately. However I suffer from idiopathic hypersomnolence (among other things) so I was hoping Fitbit would help provide some clues, perhaps, as to what is really going on. I understand it's not their job though and it's not a medical device. 

That being said, any idea about those horizontal lines? Do you see what I mean about horizontal lines, on a chart, typically meaning 'no variation, or 'no change'... That's the problem with this chart. They are not looking at a unique variable (ie. spo2). They are looking at some kind of crosstabulation to compute variation (which is a multi-dimensional variable). How is that calculated though?! To compute variation you need a baseline (variation to/from). The most southern point on the chart would mean absolutely no variation. So say I'm at 97% to start, and I stay at 97% all night. That would mean a flat horizontal line along the x axis, on the very bottom.

But in another scenario, I start at 97, then I go to 99... then the line would shoot up a bit in the Y axis. But then I go to 96. The line would still go up because despite the NUMBER decreasing, the variation is still there. So now I'm nearing the 'high variation', but then go back to 97, which is my starting point. Would the line then go back to the bottom of the chart? That would mean 'no variation' but that's not the case because I just went from 99 to 97. 

The other possibility is that your starting number is mid-y axis in the 'low' zone. So I start there at 97. Then I go to 99, so the line goes up. Then I go to 96, and the line goes down a little bit. How is it possible then that you could go below your starting point of 97? 

In short, what I'm awkwardly trying to get at is that a variance chart needs to display actual values. The y axis needs values, and the line itself will then show the variance (change between values). A y axis of 'high/ low' is unheard of. I am having trouble explaining what I mean because the chart is so nonsensical there are no words to explain what is wrong with it lol. 

Do you know, then, who at Fitbit would be able to explain this clearly? There is a possibility there is some secret behind it I am unaware of. But they need to be able to explain when questions are asked. Like I said, I asked customer support and they were completely clueless. They said I cannot speak to one of their technical analysts. So who next? (don't get me wrong, I have a life and I'm actually drowning in things to do right now... but this is too important to drop). 

I'm literally just trying to wrap my head around this. 

[Guy_]

@juliefouc- There seems to be a confusion perhaps, may be mine.

The SpO2 is a percentage Oxygen level. An average value for the night is calculated with the high and low given.

This is NOT shown in the Estimated Oxygen Variation, which only shows variation, not Oxygen levels.