Ex-Oura Advisor: The company is artificially altering your Sleep Data

Dr. Cody dives deep into Oura published methods and analytics for sleep

=============================================================
=============================================================

Episode Timestamp Notes:
The problem with a ring sensor: 0:00-2:30
Issues with Sleep Staging: 2:30-3:49
Oura Tweaking The Data: 3:49-5:52
Testing against brainwave EEG: 5:52-8:07
Concerns about physical readiness scores: 8:07-10:09

[CodyRallMD]

Sens.ai: Enter “drcody” at checkout for a $100 discount www.sens.ai

[Perqd]

Sorry but your muse headband is not the gold standard, and making a cooperation with ref link to a device that is in the same field you do critic the most popular product is shady. Also you have no studies just anacdotal evidence in comparsion to a device that IS NOT the gold standard. There are actual studies on the oura ring and the gen3 algorithm that test on 97 subjects and found the ring to be at "84% accuracy rate for sleep staging when compared to PSG" (Study from the e University of Tokyo and you can find others online)

[debbiebridygham]

I’m currently using my Watch with the AutoSleep app, my Ōura ring, and a WiThings sleep pad sensor, of these, the WiThings significantly overestimates my sleep time, while the ring significantly underestimates it, and the watch with AutoSleep seems most accurate. Your video was helpful in understanding why the ring underestimates my sleep time, though I also have a significant hand tremor which is likely related as well.

[theanswerisfortytwo]

My experience is that SPO2 and sleep stage data sucks for Apple Watch/Oura/Garmin, but actual sleep detection like hours slept works really well on the oura ring and the only thing I use it for. The thing I love about Oura vs the Apple Watch (I wear both) is that it captures the naps really well.

[johncook]

I've been Team Oura since just before the pandemic, but I've really done a flip. As a lifelong NREM-Parasomnia sufferer, I *really* want to see precisely when I'm in various sleep stages, not just data-smoothed estimates. I decided to additionally wear a FitBit Sense 2 a few months ago, and it's been notable how much the sleep cycle maps differ between the two products.
On top of all of this, I'm on my second Oura in that time period, and found that my batteries became weak at about the two year mark. My second is now getting to that point, and I don't think I'll be replacing it (it was a warranty substitution).
The Muse is fascinating, but probably a bit invasive, given the sleepwalking... Still, I'd love to see the data it generates. Have to put some money in the piggy bank!

[StealthRetail]

Your videos are getting better and better

[chriscowan]

Nice video thanks. I have a ringconn, which I probably has algorithms similar to the oura. It's excellent for recording when I fall asleep, and periods during the night when I'm awake... and if I wake early in the morning for an hour and go back to sleep it doesn't miss a beat. Like your experience with the oura, it's telling me I don't get enough rem, which I suspected was wrong, since i seem to dream quite a lot. The ring is way way better than my garmin instinct, which often "thinks" I'm alseep when I'm just lying quietly, and doesn't give me any deep sleep much of the time.

[igorchak]

What if I combine my Ring data with my Muse EEG data to get more accurate average values when I sleep?

[choc]

Any update on the new firmware update changing your sleep scores?

[indigoa]

My experience with the Oura Ring has been generally positive. The machine learning system they use compensates for many of the limitations inherent in their sensor technology. It accurately records sleep phases within expected parameters. However, the daily fitness recommendations are not as reliable and, in my opinion, lack value."

[IamShaunnaNakice]

This was a great video. I am on day 3 with my ring. Thanks so much for sharing this.

[markolson]

The Oura seems to accurately judge my sleep latency, but my beef is that they ding my score for falling asleep too fast. I’d caution against using the Muse as a reference standard until they actually incorporate the new algorithm into the device. They move so slowly! The original algorithm (which it still uses) isn’t accurate. You should try to find a used Dreem headband.

[seangrant]

Hi Cody, I love your videos. I would be curious to hear your thoughts on the Somnee sleep neurostimualtion headband. It has Matt Walker as a scientific advisor

[Voy]

Title is click bait, there is nothing bad with what they are doing. This is standard practice in building systems that use noisy inputs.
They are doing best they can, it is actually amazing that a ring on a finger can be so precise in detecting your sleep stages.
I hate their subscription BS, but nothing wrong with their approach to measurements.

[yildiraykomurcu]

Thank you for the follow up video Cody. I wrote a long comment in your comparison video, Muse vs Oura vs Whoop, you might have missed it since I didn’t see a response. I believe that all the non-EEG devices have to resort some sort of a machine learning algorithm in order to predict sleep stages. Even in this case, they may not access to N2 stage, which is characterized by sleep spindles and K-complexes and which is often the case as PPG devices only report deep and light sleep. Quantified Scientist channel made extensive reviews on wearable devices and he has come up with the conclusion that the best device that imitates an EEG device in sleep tracking is Apple Watch while Oura, Whoop and Fitbit are distant second. I respect what Oura is trying to do. However, their heart rate measurement was not among the best which would ultimately affect the sleep monitoring.

[ayemjhenzi]

Can you do a deep dive of ultra human ring?

[alyousuf]

Okay now tell us whats the most accurate consumer sleep tracker 😊

[th]

I used the Muse S for sleep tracking but the band was unreliable. Muse sent me an improved band that seems better but their multipurpose band for meditation and sleep I believe is poorly engineered from a reliability standpoint and is a weak link. It’s also frustrating when I can get the the original Muse I have from 2016 or 17 to quickly sense all sides of my brain while I have to faff about with the Muse S and it’s lateral sensors, constantly readjusting.

I use an Oura Ring Gen 3, Watch Ultra 2 and a Whoop band as well.

From a purely “how do I feel when I wake up” standpoint. I have say the Oura Ring 3 most accurately mirrors that with the Apple Watch next and the Whoop band last.

I would use the Muse S for sleep tracking if the band was reliable and less frustrating to use. It’s ironic how something that’s supposed to put you in a calm state of mind ends up being somewhat infuriating.

Be advised this is MY experience and hardly scientific.

Wishing everyone on here restful sleep !!!

[jorgebarreras]

Having a bias term in your ML algorithm is not "data massaging" or shady in any way. Practically every statistical learning tool other than rules of thumb allow you to incorporate bias terms. The observation that early in the night you are more likely to be in deep sleep is probably true and helpful for prediction, so, what is the "massaging" part? You really think that the product you are comparing against (and also partnered with, which is pretty shady actually) does not use ML algorithms that have bias terms?

This is just ignorant

[luizz_k]

Bring heart tracking to the mind monitor app!