Apple Watch iOS16 sleep app vs Autosleep app

[CapeTownMac]

I have been using Autosleep to track my sleep for a few years. It generally agrees with the simple Apple sleep measurements but, up until now, the Apple app did not give a breakdown between the various sleep stages and I used Autosleep to measure my deep sleep. According to the Autosleep app I normally get between 1 and 2 hours of deep sleep.

I used the Apple iOS16 app for the first time last night and it registered zero deep sleep compared to Autosleep which registered 1 hour, 45 minutes of deep sleep. The relevant screen shots are attached.

I'm inclined to believe Autosleep because it is an established product and I have found it to be intuitively accurate. Also, every normal human has deep sleep in a sleep cycle and it would be very unusual to have no deep sleep.

The Apple app has different categorisations. Autosleep does not measure "REM" sleep on the basis that it is not measurable without measuring your actual brainwaves but the Apple algorithm is supposed to approximate it.

If anyone else is interested in this topic and would like to compare notes, please share your experience/insights.

Thanks

 

[huddman]

I absolutely agree with your conclusions. I've used IOS 16 to track 3 nights of sleep and each night it registered zero deep sleep. On those three nights I also used the Oura version 3 ring and a Fitbit Inspire 2 both of which registered deep sleep. From my perspective Apple has some cleaning up to do with its new sleep algorithm.

 

[Apple_Robert]

It has been my experience, that AutoSleep has a tendency to overestimate a lot of the time.

Last night, Apple showed I had 33 mins of deep sleep while AutoSleep saw 2hrs 12 mins. Looking at last night's results, I am leaning towards Apple being more accurate.

 

[CapeTownMac]

I absolutely agree with your conclusions. I've used IOS 16 to track 3 nights of sleep and each night it registered zero deep sleep. On those three nights I also used the Oura version 3 ring and a Fitbit Inspire 2 both of which registered deep sleep. From my perspective Apple has some cleaning up to do with its new sleep algorithm.

I think zero deep sleep is very unlikely unless there are special conditions. "The quantified scientist" on YouTube did a scientific analysis of the iOS app in which he compared it to an EEG device and concluded it was the most accurate of the wearable algorithms but he tested the beta version, perhaps there were some last minute changes that messed up the algorithm. The average adult gets 1 to 2 hours of deep sleep per night.

 

[CapeTownMac]

It has been my experience, that AutoSleep has a tendency to overestimate a lot of the time.

Last night, Apple showed I had 33 mins of deep sleep while AutoSleep saw 2hrs 12 mins. Looking at last night's results, I am leaning towards Apple being more accurate.

The only way to be sure is to compare the results to an EEG device which I don't have access to. The average adult gets 1 to 2 hours of deep sleep per night (reducing with age) so based on your results I'd say Autosleep is more likely to be correct?

 

[Apple_Robert]

AutoSleep has been updated to reflect wrist temperature.

 

[ericg301]

I prefer the native integration. And I will tend to trust Apple and their dedicated watch team of doctors and researchers rather than a single indie developer.

But in the end, there’s still only so much data a wrist device can capture and interpret. To me it’s all about relativity anyway. Did I sleep more last night than yesterday? Do I feel more rested? Did that third cup of wine in the evening make me sleep less?

 

[OriginalAppleGuy]

Since getting the Ultra with its improved battery, I now wear my watch to bed. I've noticed some changes in how the app behaves the more I wear it. For instance - at first when I'd get up to pee around 5 AM, it would consider that my wake up time. Now, it's just awake time.

I have had a sleep schedule set up for some time now. Figured that would be enough to get wrist temperature. It wasn't, have to set the toggle for Sleep Focus to use that schedule. I resisted using Sleep Focus since I need to be available 24x7 if anything catastrophic happens (in IT). So I acquiesced but enabled some apps during sleep focus and now I'm in the process of getting wrist temp readings. Pretty bizarre to me how Apple went about this as pulse, respiratory rate and oxygen work without focus on.

Over the last week - my average deep sleep is 42 minutes. Avg REM is 1 hr 14 min.

 

[BarinCrocodile]

I have been using Autosleep to track my sleep for a few years. It generally agrees with the simple Apple sleep measurements but, up until now, the Apple app did not give a breakdown between the various sleep stages and I used Autosleep to measure my deep sleep. According to the Autosleep app I normally get between 1 and 2 hours of deep sleep.

I used the Apple iOS16 app for the first time last night and it registered zero deep sleep compared to Autosleep which registered 1 hour, 45 minutes of deep sleep. The relevant screen shots are attached.

I'm inclined to believe Autosleep because it is an established product and I have found it to be intuitively accurate. Also, every normal human has deep sleep in a sleep cycle and it would be very unusual to have no deep sleep.

The Apple app has different categorisations. Autosleep does not measure "REM" sleep on the basis that it is not measurable without measuring your actual brainwaves but the Apple algorithm is supposed to approximate it.

If anyone else is interested in this topic and would like to compare notes, please share your experience/insights.

Thanks

I have experienced the same thing with the new built in Apple Sleep app. Since tracking my sleep using the app on Apple Watch 8 the most deep sleep I have recorded in a night is just 20 minutes. Most nights it records that I have no deep sleep at all.

Essentially these apps are making approximations, and do little more than look at your pulse and movement, which they will have correlated with EEG data. While that may be sufficient for monitoring 'asleep' vs 'awake' i imagine that the results are not particularly accurate for the actual sleep stages. There is not much documentation available to assess the overall accuracy of these correlations, nor their variability among different people.

My guess is that because the watch relies mostly on pulse, the user's resting pulse probably affects the reading substantially. People with a low resting pulse while awake will have less pulse variability in sleep, and therefore defining the threshold between sleep stages will be less accurate. My resting pules while awake is 50, which is why I think the watch cannot pick up the difference between 'core' and 'deep' sleep.

In summary, they are not really precise readings and you should not lose any sleep over it.

 

[matrix07]

The only way to be sure is to compare the results to an EEG device which I don't have access to. The average adult gets 1 to 2 hours of deep sleep per night (reducing with age) so based on your results I'd say Autosleep is more likely to be correct?

 

[deijmaster]

I prefer the native integration. And I will tend to trust Apple and their dedicated watch team of doctors and researchers rather than a single indie developer.

Yes.

I've been using AutoSleep for years and since Apple integrated sleep tracking, I've noticed significant differences between the two of them. I'm not too keen to take position on what type of sleep, as I think this is very questionable, but simple sleeping period differences a very common.

Just last night, AutoSleep registered 5h51min of sleep, while Apple Health registered 7h1min; the second making more sense.

When it comes to integration, analyzing patterns, algorithms and access to specific data and API's, general trust should favour Apple. I'm sure AutoSleep is good, but update cycles for iOS are very particular and many changes come and go -> it's very hard for smaller developers to follow these, and fully understand the impacts of these.

 

[grandyb]

It's been a while since anyone posted to this thread but I just signed up and wanted to add some thoughts. I recently updated my Apple Watch to a Series 8 and began using the sleep tracking function. I initially used the AutoSleep app and got results that were consistent with my perception of my sleep. After a few weeks, I started watching the sleep stages results as well and have experienced the same result as many people posting here. Using autosleep I get typical deep sleep readings of 1 1/2 to 2 1/2 hours most nights while sleep stages consistently shows either no deep sleep or almost none (<15 minutes). Additionally, the AutoSleep result is out of sync with my perception as I usually wake up feeling rested and refreshed while sleep stages results seem to indicate otherwise. I applaud Apple's commitment to improving consumer health technology and overall I think they've done a pretty good job. They are, however, far from perfect and this technology is a work in progress. I would caution anyone from assuming that because it's from Apple it must be good. As an example, I've been using Apple Watch to moniter ECG's for a couple of years now as I have a family history of April Fibrillation. My apple watch series 4 regularly characterized my ECG as atrial fibrillation. This, however, was in complete disagreement with clinical ECG results that I received using a 12 point hCG at my cardiologist's office. In addition when he analyzed ECG's that were produced by the Apple Watch indicating AF, he was emphatic that what was showing was not AF. I had many conversations with the Apple health team,where they insisted that there were no problems with the accuracy of the watch for that purpose. Approximately a year into this, I finally got an admission that there were certain rhythms that the Apple Watch misdiagnosed as AF, resulting in false positives, and that was happening tro me. I believe AutoSleep is most likely closer to reality given that they have had much more time to refine their algorithm. All of these consumer facing tools are works in progress. I'm convinced that the sleep stages algorithm is at least somewhat broken. I expect it will improve over time but I'm pretty confident it's not accurate at this point, at least for some people.

 

[broncopde]

…I initially used the AutoSleep app and got results that were consistent with my perception of my sleep. After a few weeks, I started watching the sleep stages results as well and have experienced the same result as many people posting here. Using autosleep I get typical deep sleep readings of 1 1/2 to 2 1/2 hours most nights while sleep stages consistently shows either no deep sleep or almost none (<15 minutes). Additionally, the AutoSleep result is out of sync with my perception as I usually wake up feeling rested and refreshed while sleep stages results seem to indicate otherwise.

Should this last sentence be “Additionally, the Apple Health Sleep Stages result…”? Thanks!

 

[grandyb]

Apple Health Sleep Stages result

Should this last sentence be “Additionally, the Apple Health Sleep Stages result…”? Thanks!

Yes, thanks, I meant to say the Apple Sleep Stages result is out of sync with my perception as I usually wake up feeling rested and refreshed while sleep stages results seem to indicate otherwise.

 

[steviehype]

I'd like to fire up this conversation again because I'm struggling to find if people are accounting for heart rate dip.

I installed Autosleep recently because, as mentioned above, Apple Sleep is recording barely any deep sleep for me.

Autosleep had similar results to Apple Sleep until I found a setting for heart rate dip. Looking at previous data my heart rate can dip around 20% at night. I consider myself fairly fit, I run regularly and my heart patterns for my age fit a range often called "athlete". So when I found a dip setting in "Sleep Goals" within Autosleep it made sense to tweak it to "Athlete". (screenshot attached)

This has dramatically changed how much Deep Sleep is recorded and aligns better to how I feel.

To the folk using Autosleep, is this your experience as well?

 

[grandyb]

I'd like to fire up this conversation again because I'm struggling to find if people are accounting for heart rate dip.

I installed Autosleep recently because, as mentioned above, Apple Sleep is recording barely any deep sleep for me.

Autosleep had similar results to Apple Sleep until I found a setting for heart rate dip. Looking at previous data my heart rate can dip around 20% at night. I consider myself fairly fit, I run regularly and my heart patterns for my age fit a range often called "athlete". So when I found a dip setting in "Sleep Goals" within Autosleep it made sense to tweak it to "Athlete". (screenshot attached)

This has dramatically changed how much Deep Sleep is recorded and aligns better to how I feel.

To the folk using Autosleep, is this your experience as well?

I continue to believe there is a problem with the Apple Stages algorithm and the reported Sleep Stages data does not match my perception. As I mentioned above, I have the same result of seeing a large difference in the deep sleep reported by Sleep Stages (usual almost none) and AutoSleep (typicall around 3 hours). The Apple Stages data is also far different of my perception of my sleep quality while the AutoSleep data is highly consistent with my perception.

I want to make sure I understand your point about the HR drop. Was your initial data for deep sleep low in both apps? It sounds like the tweak you made to the setting for HR drop would have resulted in a change to the AutoSleep data. I am also aerobically and cardiovascularly healthy and train regularly for endurance events. Did the adjustment you made in AutoSleep reduce the reported deep sleep value in the data reported by AutoSleep? I changes the seting you suggest and will be interested to see my result. I will post an update when I have some data. Thanks!

 

[steviehype]

The change I made increased the amount of deep sleep recorded by AutoSleep. This fits my perception of my sleep better, it aligns with how I feel.

So I feel AutoSleep is now more accurate that Apple Sleep for me.

I’d be interested to hear how it goes for you 👍🏼