Last Updated on February 15, 2024 by Art

(Sleep #1 of 3) For Present and Future Health…

Sleep sensing devices and apps are being heavily marketed. The target audience appears to be people who are dissatisfied with their sleep, plus people who want to improve their present and future health.

Today’s blog gives a summary of sleep sensing options that are available. The next blog will explain what sleep does, and how it may relate to Alzheimer’s Disease and other forms of dementia. The third blog of this set will tell you what I learned from one sleep app during many months of use.

Sleep Sensing: a Glut of Choices

Glut is the word to describe sleep sensing options today. When I search Sleep in my iPhone App Store, I count 239 apps! And no matter what search approach you take, you are sure to receive Too Much Information.

No one tries to compare all of them, but it’s easy to find some “top ten” lists. Perhaps the best is Jen O’Brien’s in CBC.ca. However, also worth reading are those by Sleep Foundation, Forbes and Good Housekeeping. O’Brien cuts through the tangle by defining three useful app categories:

• Apps offering sounds to help you sleep
• Those that soothe and clear the mind
• And sleep trackers that provide data in order to identify and address issues

We should read all such articles with caution, because the authors generally make money on click-throughs links they provide. That conflict of interest may be why most reviews are positive overall, even when they point out shortcomings in a product.

Here’s how I sort through the noise: as a scientist, I want to see data, not someone’s squishy opinion. And if I want to hear sounds or soothe my mind, I don’t need an app to guide me. So I begin with a preference for the last of the three types, those that provide concrete sleep data.

            Data Tracker Comparisons

Within the data tracker category, I’m most interested in accuracy, initial cost, ongoing cost and intrusiveness. Let’s start with how well the sleep trackers function.

I searched for research reports comparing the sleep trackers and immediately ran into a problem. Published articles require the authors to reveal any conflicts of interest, and in this area the conflicts are numerous: in most cases the authors work for the company that sells the tracker, and in other cases the company paid for the research report.

As a test, I e-mailed the author of a study comparing the Oura ring to polysomnography. The article states that their data covers users ranging from 15 to 73 years of age. So I asked the author how the performance of the device compares for users of different ages. He replied that they had not looked at that parameter. And that moreover, he could not reveal any information not in the published article because all of the data belongs to Oura. He advised me to contact Oura, which I did, and I never received a reply.

            And the Research Shows…

I only found three research articles that appear to have no significant conflicts of interest:

• Miller et al, A Validation of Six Wearable Devices…, August 2022. This article compares polysomnography, Apple Watch, Garmin, Polar, Oura, WHOOP and Somfit. Although WHOOP helps support the author’s laboratory, at least the comparisons involving other devices are presumably objective.
• Chinoy et al, Performance of seven consumer sleep-tracking devices…, December 2020. This article compares polysomnography, Actiwatch, Readiband, Fitbit Alta, Garmin Fenix, Garmin Vivosmart, EarlySense, ResMed S+ and SleepScore Max.
• Lee et al, Validation of a Consumer Sleep Wearable Device…, September 2019. Compares polysomnography with Fitbit Alta and Actiwatch.

What are the results of these third-party comparisons? Generally, that sleep sensing devices all perform equally well, and equally poorly. They are about 67% accurate in noting when you change your sleep state (light sleep, deep sleep, REM sleep). However, once they pick up on that, they are 80 to 90% accurate in correctly identifying which state you are in. To clarify, here’s a comparison of polysomnography (upper track) with the Oura ring prediction (lower track):

Figure 13 from Altini and Kinnunen(https://www.mdpi.com/1424-8220/21/13/4302)

The way I read this comparison is that the ring mostly agrees with PSG, but hops in and out of its predictions, which keeps it from getting credit for high accuracy.

Remarkably, the sleep sensing devices that touch your body do not perform any better than those which are totally non-contact. Apparently, once you no longer capture the full brainwave pattern in polysomnography, you can’t really make up for that absence no matter how many sensors you have measuring motion, skin temperature, blood oxygen and pulse. A noncontacting sensor, by capturing many subtleties of body motion and position change, does just as well.

            The Importance of Intrusiveness

As I stated above, we should judge data trackers by accuracy, initial cost, ongoing cost and intrusiveness. The above discussion has covered accuracy, and device costs are easy enough to compare. But I assert that intrusiveness is a very important factor to consider.

Why? Let me offer an example: obstructive sleep apnea (OSA). OSA greatly increases risk of stroke and heart failure and shortens life expectancy. So you’d think that people with that diagnosis would go to some trouble to correct it. However, the most widely prescribed treatment, CPAP, is refused by one-third of patients, and of those who try it, many drop it within one year. They abandon CPAP because sleeping with a pressure mask is just too much bother, even to save one’s life.

If physical inconvenience won’t persuade many people to save their lives, it’s even less likely that they will tolerate sleep tracking if it’s annoying. After all, sleep tracking is a nice-to-have, not a got-to-have, and its benefits, if any, are gained only over time. Therefore a sleep tracker should be minimally intrusive to have a chance to deliver useful results.

Most of us do care about our sleep. Getting a good night’s sleep facilitates every activity of life. And, more ominously, deficits in one or another sleep measurement may foretell dementia in later life.

With that warm-up, let’s talk about sleep sensing options, with one eye on their intrusiveness.

Maximum Annoyance: In-Lab Polysomnography

Test subject wired for polysomnography

Polysomnography (“multi-parameter sleep measurement”, or PSG) is, hands down, the most intrusive approach to sleep sensing. However, it is also the “gold standard,” meaning the only technique accepted as correctly classifying all phases of sleep.

The “gold standard” is not completely golden, however. The American Academy of Sleep Medicine offers training programs to improve the reliability of PSG ratings. However, a 2022 review of published studies found only 76% agreement in the sleep scores found by different sleep technicians; and the accuracy with which deep sleep and REM sleep were identified was only 57% and 69% respectively. Therefore, the fact quoted above that sleep sensors in general agree with PSG 67% of the time is not too shabby, considering that trained PGS technicians also disagree with one another about that often!

Even this degree of accuracy comes at a price. The patient visits a sleep lab, where technicians attach electrodes to scalp, temples, chest and legs. Together, these monitor brain waves, eye movements, heart rate, breathing pattern, blood oxygen level, body position, chest and abdominal movement, limb movement, snoring and other noises. The technician is present and can verify that the sensor data is mutually consistent. This is very important when diagnosing serious issues such as sleep apnea. PSG requires a night sleeping in a clinical lab and costs about $3,000, usually covered by insurance.

I should point out that none of the alternatives to PSG are reliable, nor certified, for diagnosing sleep apnea. (A researcher comments that the self-test devices “work well for those with normal sleep and less so for those with abnormal sleep.”) Therefore, if you have any apnea symptoms, or your sleep partner tells you that you do, do not waste time with self-diagnosis but consult your physician at once.

At-Home Polysomnography

The next most intrusive option is in-home polysomnography, typically costing $600, also usually covered by insurance. Patients typically affix just a finger oxygen probe, a chest belt, and a nasal tube before going to sleep. In some cases there is also a microphone.

Unfortunately, the in-home setup does not include brain wave sensors, which need to be carefully positioned by a trained technician to provide reliable data. Therefore, stages of sleep have to be inferred from other measurements, making this approach distinctly less accurate than sleep lab tests.

Wearable Sleep Sensing

Once you step down from sleep lab tests to self-administered tests, the intrusiveness drops significantly, but tests still retain useful accuracy. However, this accuracy comes at a price.

            Muse Headband

The wearable sensor that comes closest to measuring brain waves is the Muse S. You might also call it the most dorky of the wearable devices. However, dorky doesn’t really count because most users won’t wear it all through the day. Muse is intended to be an aid to meditation and a monitor while sleeping. In addition to brain waves, its sensors measure blood oxygen, acceleration and orientation.

Unfortunately, reviewers complain that the headband is difficult to get positioned properly and uncomfortable to wear overnight. The headband costs $399. Moreover, if you want full access to their 500 meditation routines, you’ll have to pay for a subscription.

            Apple Watch

Apple Watch Ultra (image by KKPPCW)

Unlike the Muse headband, the Apple Watch is something you can wear at all times without looking like a refugee from a research clinic. Although it was not invented as a health monitor, it has multiple functions. These allows the watch to measure step count, heart rate, irregular hearth rhythm, cardio fitness, blood oxygen level and fall detection. To obtain all these functions, you need a series 6 or later watch. An Apple Watch is likely to cost you $250 or more.

            Oura Ring

Oura Ring Gen 3 (image by Tpnkl)

Here’s yet another sleep sensing device, one which is probably even less noticeable than a watch. The Oura ring is a bulky-looking ring that some might call ugly, and which costs $300 to $550.

It’s packed full of sensors, which measure heart rate variability (HRV), blood oxygen, body temperature and sleep duration. Reviewers say that all the useful data is locked behind a paywall requiring you to pay $72 per year for a subscription. They also are worried that it’s easy to lose.

Non-Contact Sleep Sensing

All these clever sleep sensing devices are designed to drain your bank account and make someone else rich. However, they are not the only game around. You might be surprised that they have serious competition from devices that never touch your body!

Here are some non-contact sleep sensors to consider:

            Snore Lab App

I tested SnoreLab in February 2023. At that time it was one of the highest-rated apps on the Apple App Store. It is not a device: it uses the microphone built into your smartphone to record and then score your sleep and snoring.

The free version of this app is so limited that it’s practically useless. It captures only four sequences, of the loudest noises. Moreover, after your first 5 trials, it forces a 24 hour dead time between trials; thus you cannot take two measurements during one night, to compare sleeping positions. In addition, the free app stores data for only 3 sessions.

The paid version of this app charges $5.99 per month or $23.99 per year. It offers full night recording (which your phone can already do with Voice Memos) and unlimited data storage, both features in which I don’t see value.

My free trial was enough to highlight serious deficiencies in this device:

• Because the app detects all sound, it’s not useful when there are two sleepers in the room, such as me and Nola. Since we don’t plan to use separate bedrooms, this app can’t advise either one of us on how to improve our sleep.
• Another flaw is that the app processing does not detect gasping for breath, which is a revealing symptom of the serious medical condition of obstructive sleep apnea.

It also appeared to me that the snoring data and score were inconsistent from one night to another.

            SleepScore Max

SleepScore Max (image by Pottery Barn)

Another sleep sensing approach that is non-contact is SleepScore Max. It’s a product of ResMed and is a newer model of SleepScore S+, which some researchers tested.

The Max is basically a radar set and computer that sits next to your bed. The device transmits short radio wave pulses and receives their reflections from your body. The signal processing measures both large and small movements, and only processes return signals from people (and objects) within 3.9 feet. Thus it can sense a single sleeper even if there’s a second sleeper also nearby.

The manufacturer states that the device transmits “ultra-low power radio waves (1/10 of Bluetooth®)”. A review says that it transmits radio wave pulses at 10.5 GHz frequency. The comparison with Bluetooth suggests that the power level is about 0.25 milliwatts.

This device has been selling for $149, but Pottery Barn is currently selling it on clearance for $73.99. This price reduction suggests to me that the company plans to discontinue this product in favor of their SleepScore sonar device, discussed next. ResMed has supported the Max without a subscription charge but perhaps it’s a better value proposition for them to offer the sonar option and hope to sign most users up for a subscription.

I have not tested the Max, nor do I plan to. In modern life we are exposed to radio waves everywhere we go, and if they were very dangerous we would not still exist. Nevertheless, I see no need to increase my exposure when the alternative is sound waves, which humans have evolved to tolerate.

            SleepScore

SleepScore, also from ResMed, does away with the radar set on your nightstand. Instead, it uses the speaker and microphone in your smartphone to measure your body movements using sonar. The sonar frequencies used are around 20 kHz. These frequencies are at the very top limit of human hearing, and far beyond the top for those who have lost some of their hearing with age, as I have.

Since humans evolved surrounded by a wide range of natural sounds, we can expect that a sonar-based device does not pose health risks.

SleepScore, like all the app providers, would like to sell you a subscription for $6 per month or $50 per year. Although they promise personalized help with your sleep problems, reviewers report that the support they provide is just a collection of articles to read. I have found the app pretty useful in its free configuration and that’s how I have used it. I have used the SleepScore app for most of a year, and a later blog will describe my experience with it.

If you would like help with your sleep, or better sleep, it’s worth trying some of the available apps. Unless you just like to own lots of tech, you don’t need to buy a device that you don’t already own. A few trials will show you what outputs are available, and whether they’re useful to you. The next blog describes how sleep health relates to later dementia, and the last of this series tells my experience test-driving one particular sleep app.

Image Credits:
– Lead image compiled from vectorscape, Firkin, pnx, user unknown on openclipart and Veterans Administration
-Compare polysomnography with Oura ring, figure 13 from Altini and Kinnunen
– Polysomnography wiring from Veterans Administration
– Apple Watch Ultra by KKPPCW
– Oura Ring Gen 3 by Tpnkl
– SleepScore Max, image from Pottery Barn