Smart phones and smart watches now enable people to monitor and quantify all manner of information about their daily habits that were not possible before. Ever since I was gifted an Apple watch by my daughter, I now know how many steps I have taken, how many calories I have burned, how much exercise I have done, as well as my heart rate, respiratory rate, and so forth. While I find all that information mildly interesting, I can see how for people who are worriers or outright hypochondriacs, this can feed their anxieties.
One such item that is measured is sleep. My phone that is synced to my watch tells me each morning the quality of my sleep during the night, such as when I fell asleep, when I woke up, how many hours I slept, how many times I woke up, how many hours were spent is REM sleep, core sleep and deep sleep, and so on. In general, I have no problem falling asleep or getting a lot of sleep each night and my watch is clearly proud of my achievements in this area because it keeps congratulating me on what a great job I am doing.
There has been a lot of media attention recently to what is seen as a major problem in that people are supposedly not getting enough sleep or that the quality of their sleep is not good and that this could affect their overall health. People are being encouraged to to get more and better sleep. But now warnings are being issued that this trend may be going too far giving rise to something called orthosomnia, which is “the medical term for an unhealthy obsession with attaining perfect sleep, usually driven by a wearable device.” It appears that monitoring one’s sleep too closely and trying too hard to get better sleep may end up making the problem worse.
The term was coined by US researchers in a study published in the Journal of Clinical Sleep Medicine, which showed that, ironically, geeking out over your nightly sleep data may actually make your insomnia worse. The scientists observed how some people spent excessive time in bed trying to perfect their score, while others experienced anxiety about not achieving a good enough sleep performance.
…Katie Fischer, a behavioural sleep therapist, says that many of her clients are chasing the “perfect” night’s sleep. “The thing is, if you ask good sleepers what they do, they will usually say ‘nothing’,” she says. “They’re not thinking about sleep. They’re tired, go to bed and drift off. They wake up feeling refreshed and don’t worry about the occasional bad night because they know that happens sometimes. Meanwhile, people who come to me with problems are often putting pressure on themselves to attain a magic number of sleep hours, typically eight, without understanding that they might not need that much.”
She has seen clients distraught because their trackers have recorded only 11% deep sleep a night. “I tell them that’s pretty normal. We sleep in cycles of roughly 90 minutes and we transition between light, deep and REM phases. For most people, deep sleep will only account for 13-23% of the night. Time awake is also a normal part of sleep. We can have between three and six rousings a night; often we’re not even aware of them. Most people have no idea what normal sleep looks like, so they panic when they see stats with lots of awakenings and not much deep sleep.”
I have always known that I wake up and drift back to sleep several times in the night. When I first got my watch, I was interested to see that pattern being reflected in the data but it did not surprise or bother me. I assumed that it was like that for everyone, and apparently it is. But it seems like some people think they should have uninterrupted sleep and that the lack of it is a cause for concern and obsess about it. That may be the real cause for concern.
Part of the problem is that all this data may not be actually useful in changing things.
Dr Neil Stanley, the author of How to Sleep Well, has been working in the field of sleep research for 42 years, starting out at the neurosciences division of the RAF.
…Stanley also wants to raise a more fundamental question. “I’ve spent my career hooking people up to monitors in spaces designed to be the finest bedrooms on the planet. I’ve sat there watching the squiggly lines that measure the different stages of sleep. And, to be honest, it doesn’t mean a thing. Because what even the most accurate data cannot do is influence you to sleep more hours. So, for the general public, what’s the point?”
He also says that if there is one night when someone misses out on the sleep they need, the body will usually rebound to catch up the night after. “I’m not saying that you can’t improve the quality of your sleep; there are lots of things you can do. I would recommend anyone to keep a pen-and-paper sleep diary of what time they go to bed, how tired they feel, when they wake up and how awake they feel at 11am. In other words – listen to your body, not data.”
But the desire for data is insidious and there are companies that are ready to feed the craving.
A recent article published by Loughborough University on the future of sleep tech mooted concepts including pyjamas embedded with sensors to track changes to posture, breathing and heart rate, and hugging a robot pillow coded with a respiratory pattern to help people get to the land of nod more quickly.
I know people who actually cannot sleep for more than three or four hours or so each night and that can be a real problem. But it seems like most people worry more about their sleep than it actually warrants.
I would also question the accuracy of a lot of this data. I gave up on the step-tracking on my phone very quickly because it was blatantly obvious that the data was garbage. But that’s a easy one to confirm -- “oh, hey, my phone says I’ve done 500 steps while I’ve been sitting on the couch scrolling Facebook!” But for stuff that you can’t easily cross-check, there’s really no telling.
This is related to the fact that people often confuse precision for accuracy. They assume that if the read-out goes to 2 decimal places, then it must be that accurate, when in fact it’s very often not. A good example of this is cheap thermometers -- they may read to 1 decimal place, but they’re often only actually calibrated to within a couple of degrees, at best. For example, I have a bunch of identical thermo-hygrometers around the house, which I calibrated against a half-decent thermometer when I bought them, and the offsets on them range between 1.8 and 2.5 degrees C. And I haven’t ever got around to re-calibrating them to check how stable they are…
Gathering data is only useful if you can do something effective with it. Several years ago, one of my friends and training partners (distance running) told me about an offer where we could get a free initial test to determine our lactate thresholds (the exercise point at which lactate begins accumulating in the blood faster than you can clear it, a marker of fitness). My reaction was “Why bother?” Endurance athletes tend to obsess over parameters such as VO2max, hematocrit, etc., but unless you’re getting regular testing and comparing those markers against your training to see what workouts/cycles are most effective for you, it’s just a number. If my VO2max is higher than someone else’s, that doesn’t mean I’m necessarily faster unless all other factors are identical, including your mental state. If it did, there’d never be races: we’d just show up with the results of our blood tests, and the person with the best numbers would be declared winner.
Obsessing over the numbers just leads to anxiety, and that anxiety can reduce performance. I am picturing “competitive sleepers”, each trying their hardest to fall asleep as quickly as they can. What a ridiculous image.
To expand a little bit on Dunc’s comment @1, I find thinking/learning how some tool is taking a measurement often more interesting than the measurements made. And, as you say, relying on a measurement without knowing how the measurement was taken can lead to serious mistakes.
So, how does a watch count how many steps a person takes? It cannot be a direct measurement, the watch isn’t watching a person walk. So the watch must be using surrogate data to conclude that A) the person is standing, B) the person is moving, C) the person is moving by moving their legs. What do most people do when the walk? They swing their arms, some people do it a lot, some only do it a little bit, but it’s how we bipedal primates keep out balance. You can walk without swinging your arms, but it’s harder than you think and often requires conscious effort to prevent your arms from swinging. Now, assuming you are a programmer trying to find some surrogate measurement to use to detect footsteps in a watch, a wrist-mounted device. Putting inertial measurement devices within the watch will tell you if a person is swinging their arms, and if the watch has several of them, they can even make a good guess when the person is standing up. There are solid-state inertial measurement integrated circuits which have 3-axis flexing beams which are quite sensitive (In my line of work I’ve seen them used in airbag designs). A programmer can take the output of one of these ICs and use that as surrogate data for counting steps. If they are good programmers they might even be able to usually tell if a person is standing, sitting, or ling down. If they are less experienced, or careless and under pressure, they may not bother. After all, who is going to check? It is very, very common in testing that verification that the product works in the proper setting (in this case walking) is all which is tested, the possibility that it would generate erroneous data in other settings (possibly in the case of sitting and swiping a tablet) is not considered.
In the end, if you don’t know how a measurement was taken, you should treat that measurement with a certain amount of caution. If it matches what is expected, it’s probably good. If it is reporting something which doesn’t look right, or even if you have no knowledge of what, roughly, the results should be, be wary about the measurement and look for collaborating data. That doesn’t mean the measurement is wrong, but the chances are higher that it is.
To return to the OP. I do not know how a watch would measure REM sleep directly. I suspect it would be from the cessation of arm movements, or possibly through pulse rate, but is that a good surrogate for identifying REM? It may be, I don’t know. Since I don’t know, I would be cautious about what the watch is reporting. Without understanding the correlation of the surrogate date to the reported data, at best it could be seen as an indicator, at worst it could be completely misleading.
To sleep better, take off your
shacklewatch before you go to bed.As a long time user of several generations of Fitbits and a couple of generations of the Oura ring, these devices can be useful, sometimes insightful, but always a sense of proportion is needed. I prefer the Fitbit for activity tracking -- it measures both some form of arm swinging (not any arm movement, it can be hard to force it to add a few extra counts) and pulse, so I get an idea of the intensity of my walking/running, and the current one shows me my pulse on the fly, so I can decide on goals both in terms of step number and in terms of minutes of activity. The Oura only showed information on the phone, after downloading it from the ring and processing it, so it was more annoying to use to find out how I am doing at the moment. Also, it so happened that I could only get a good fit for the index finger of my dominant hand, which meant chopping vegetables added to my step counts. (References to the Oura are in the past tense because now that they require a monthly subscription in order to get the processed data I gave up on them. My wife still uses an Oura ring, and only downloads the raw data and extracts from it what she wants for her purposes.)
For sleep, the Oura was better when I could access the full data -- the patterns of pulse changes over the course of each night were interesting, as were records of movement (rare for me) and moments of poor oxygenation (extremely rare), but the less detailed sleep information from the Fitbit is good enough.
Some examples: It was fascinating to see my resting heart rate go down when I started running (and then get messed up when I needed to take a steroid for a while to treat an inflammatory condition). Also, I had no idea that going on a snacking binge raised my resting heart rate and lowered my heart rate variability. (That is true even if the binge occurs around noon and I stop eating entirely several hours before bedtime.)
My wife keeps detailed logs of her data, and finds patterns that help her understand various health issues, especially onset and resolution of inflammatory conditions.
Some changes we introduced for better sleep that got reflected in the data: better curtains. Changing blanket arrangement -- your body temperature is supposed to drop a bit during sleep. Doing a couple of minutes of breathing exercises at bedtime raises my heart rate variability significantly. (My wife has a very high HRV, which is much better than expected for her age, mine was extremely low and unhealthy, with these exercises I get to the normal range for my age.)