DIY Insight from Band data!

[anon(8555314)]

I have sleep apnea and wanted to gain more insight into my sleep than what I am getting. I came to the realization that I could manually enter data into Excel from MS Health, recorded by my Band, to answer some questions that I have. For example:

Do I sleep better on certain days of the week than others?



(By the way, I typically work Sat & Sun and take off Wed and Thu, so this pattern kind of makes sense.)

I was also wondering if I slept better the day after exercising. For this one the spreadsheet classifies each of my exercise sessions by ave HR into Zone 1 - Zone 5 with my user defined parameters. I don't have very many zone 1 exercise sessions, so the sample size is small. It turns out that I do indeed get more restful sleep after exercising in Zone 2 or above:



I also wondered how much I was hurting myself by staying up late. I was hoping it didn't matter. Now I know:



I was also curious about the new sleep restoration classification that popped up on the recent update. My sleep restorations have been classified as moderate, good, or optimal. It turns out that amount of restful sleep and resting HR both vary a lot with these classifications:




I have a lot more of these, esp. showing sleep and exercise performance over time, but I still have not entered all the data to the present. All of these insights (and many more) could easily be added to MS's dashboard in the future, because all of the data came from there, after collection by my Band. All I did was enter it into excel and use formulas to automatically calculate a number of additional fields such as DOW, exercise zone, etc.

 

[defante]

Pretty cool that you can get this kind of data. Are you manually entering this data, i.e. going to each session and getting the info from there? The web dashboard now has an option to export your data to a CSV or Excel file.

 

[kenjancef]

Well done!

 

[anon(8555314)]

Pretty cool that you can get this kind of data. Are you manually entering this data, i.e. going to each session and getting the info from there? The web dashboard now has an option to export your data to a CSV or Excel file.

I wasn't aware that the data could be exported. I will have to look at that. I was entering one line for each date, just a small subset of what is there.

One of the things I have learned is that I need to get a better understanding of what constitutes quality sleep. I was primarily looking at the length f restful sleep until I started doing this, but I had one session of only moderate restoration with 2:45 of restful sleep, much more than my ave, so there's a lot more to consider.

 

[mhc48]

Looks like the nucleus for a promising new app you should look into creating.

 

[DrFix82]

Anybody could explain me what prevents my sleep to be optimal? It is classified as "Good".
I have 98% efficiency, 44 Resting HR, only 1x wakeup....

what do they expect to be optimal? 100%, 10 HR, no wake ups... 0 seconds to fall asleep?

 

[kenjancef]

yea, as soon as your head hits the pillow you should be right to sleep... lol...

Funny though...

 

[Nate Silver]

Interesting stuff Dave, and a good example of ways to grow the Band's ecosystem! I'm wishing I weren't such an Excel klutz.

 

[anon(8555314)]

View attachment 103439

Anybody could explain me what prevents my sleep to be optimal? It is classified as "Good".
I have 98% efficiency, 44 Resting HR, only 1x wakeup....

what do they expect to be optimal? 100%, 10 HR, no wake ups... 0 seconds to fall asleep?

I think that is an excellent question, and I wish MS Health would provide more in-depth info for those who want a more elaborate explanation. I suspect that "awake" isn't necessarily awake but may be part of the REM cycle, but I am really not sure. And I also wonder how it deals with individual differences. Does it primarily compare my sleep last night to my previous data, and base it on my comparison, or does it compare it to some objective standard?

Here is one example of what was for me considered optimal sleep:

 

[Talderon]

Thanks to your post, I did some digging and found this: http://forums.windowscentral.com/mi...ndows-desktop-application-microsoft-band.html

Something to check out. I hope to have some stuff done this weekend to more or less automate importing of the data this application spits out.

 

[stephen_az]

View attachment 103439

Anybody could explain me what prevents my sleep to be optimal? It is classified as "Good".
I have 98% efficiency, 44 Resting HR, only 1x wakeup....

what do they expect to be optimal? 100%, 10 HR, no wake ups... 0 seconds to fall asleep?

I am sorry but you are asking end users of a device to compare/explain sleep cycles and explain things they lack the training to address. Unless there is a doctor lurking here who is a sleep specialist, and who doesn't grasp the concept appropriate places to give advice, there is not a single useful insight you can get on the subject here. It is also really not a competition nor should you worry about whether your sleep is optimal or good. This is a danger of this sort of device. People take data they do not understand and attempt to use it as the basis for health decisions instead of as information to help with those decisions. They also then assume you can bypass the medical professionals who need to be part of the equation.

 

[stephen_az]

I think that is an excellent question, and I wish MS Health would provide more in-depth info for those who want a more elaborate explanation. I suspect that "awake" isn't necessarily awake but may be part of the REM cycle, but I am really not sure. And I also wonder how it deals with individual differences. Does it primarily compare my sleep last night to my previous data, and base it on my comparison, or does it compare it to some objective standard?

Here is one example of what was for me considered optimal sleep:

View attachment 103442

The insight can be gotten by taking the data and talking to your doctor. I seriously doubt you will ever see anything more refined (beyond data enhancements) because Microsoft would be setting itself up for perpetual litigation for wrongful deaths, injuries, etc., if they were to provide what amounts to canned medical advice as part of the package. It is the user's responsibility to take the data and use it within an appropriate context for their health and wellbeing needs....

 

[William Grafton1]

David, you should seriously consider creating a Connected App for this. It looks like you've done a lot of the work for it already.

 

[DrFix82]

Dear stephen_az, you keep saying to talk with a doctor.
Here nobody is making any medical speculation or thinking. I'm only interested in how the MS algorithm calculates the "quality". I'm not thinking of having any medical trouble nor I want to talk about it and reading all the posts seems that nobody here is doing it.

 

[NBrookus]

I suspect the app is doing some sort of running calculation on sleep restoration. For one, that would make sense: one good night of sleep doesn't erase weeks of poor sleep. And it's actually pretty hard to "catch up" on sleep.

Second data point:


I most definitely have sleep debt right now. Although I got a pretty good night's sleep, I'd say the overall assessment is correct.

The note about improving restoration by getting more activity is a bit cryptic. I'm not sure if it's suggesting activity gives you something to restore from (which makes little sense), or if it is alluding to the fact you will generally sleep better if you exercise and is just worded poorly.

 

[gadgetrants]

The note about improving restoration by getting more activity is a bit cryptic. I'm not sure if it's suggesting activity gives you something to restore from (which makes little sense), or if it is alluding to the fact you will generally sleep better if you exercise and is just worded poorly.

I read it as: if you run, you will be tired...if tired, you will sleep more. :smile:

-Matt

 

[NBrookus]

I read it as: if you run, you will be tired...if tired, you will sleep more. :smile:

Problem is, being tired and wanting to sleep often has very little to do with whether or not you CAN sleep.

It's also possible that they are using some sort of sensor reading that don't appear in the graph. For example, maybe the readings of the movement counter on a more granular level than the light/restful threshold. There has to be something they are using that doesn't show on the graphs. Otherwise why is last night "low," but this day from last month "good"?





Honestly, without knowing *why* it says that, it's impossible to use their analysis to make positive changes, which is what these gizmos are supposed to help us do. Their analysis may also be flawed. Certainly if you took your graphs to the doctor as Stephen suggests, she'd never use this to make any sort of diagnosis; if needed she'd send you for a sleep study.

Matt, didn't you say your sleep was terrible? What's your restoration level usually at?

 

[gadgetrants]

On average I'm getting ~1 hour of "restful" a night. I've happily noticed a growing trend of 1.5 to 2-hours, though those are still outliers. It's probably placebo but I imagine I feel a bit more "reenergized" any time I get past an hour of restful (according to the Band). Here is what my last month looks like...pretty much a lot of "good" restoration with some "moderates" mixed in:



\BeginReallyGrouchyComment

BTW I totally agree with you that the term "restorative/restoration" leaves a really bad taste in my mouth. I think it's laughable to argue the "algorithm is proprietary" and to not disclose how each measure is computed. I don't trust Microsoft to protect my health...a wink and a nod is not enough. It's totally OK if there's secret sauce cooked into the sensor engineering. It's not OK that we sit here hopelessly trying to reverse engineer their dashboard. I already ranted (extensively) that putting "resting HR" on the sleep page causes uneccessary confusion. An undefined "sleep restoration" measure is equally bad.

\EndReallyGrouchyComment

​-Matt

 

[NBrookus]

Yeah, I don't mind the secret sauce. I do find it a problem that the "observation" they are making seems to have no correlation to the data they are showing us.

Wasn't there a feedback link on the web dashboard? If so, it's gone now.