Tech Longreads

Sleep tracking on a smartwatch. Can you trust wearables?

sleep with smartwatch

Only doctors and people suffering from severe sleep disorders knew about sleep stages a decade ago. Nowadays, any fitness tracker user can easily analyze their sleep with all its stages and quality evaluation. But it didn’t make them sleep any better.

The popularity and availability of fitness trackers has showered users with a huge amount of physiological metrics that they don’t know how to deal use. Some of them are quite important but often ignored (e.g., resting heart rate), while others, by contrast, should be taken a bit easier. And sleep stages analysis is one such example.

If you enjoy checking sleep reports prepared by your fitness bracelet, can you answer the following questions with confidence:

  • How does it improve your sleep quality?
  • What should you do if the fitness tracker shows a lack/excess of some sleep stages? What causes it?
  • What affects the frequency, number, and order of sleep phases?
  • Are sleep cycles the same for all people, and can you identify any pathology using a wristband report?

And the main question is whether we can trust the beautiful graphs drawn by a $50 gadget when complex devices that cost thousands of dollars are used in medicine to analyze sleep? Moreover, it takes a lot of knowledge to understand the measurement results.

We’ll try to deal with these questions and learn how to analyze sleep graphs generated by various fitness trackers. Also, we’ll find out whether they can improve your sleep quality somehow.

The sleep structure. Or what are the phases of sleep?

Sleep is more complex than it may appear at first glance. We don’t just feel drowsy, then close our eyes and slowly “switch off ” to come out of this state the next morning.

Sleep has a well-defined structure. It consists of 4 to 6 segments (cycles), each between 70 and 110 minutes long. The brain constantly repeats these cycles from the moment you fall asleep until you are fully awake. If you slept 6 hours, you had to have 4 cycles (~90 minutes each on average), and if you slept 8 hours, you had to have about five cycles.

Each cycle itself consists of phases or stages, and there are four of them. These are the sleep phases that smartwatches track (except mostly the first phase):

StageNameDescription
1NREM1Drowsiness
2NREM2Light sleep
3NREM3Deep sleep
4REMParadoxical sleep

Let’s see what an actual professional sleep record (hypnogram) looks like:

how does sleep structure look

The horizontal line shows the time of day (from 10 pm to 8 am). The vertical line indicates the sleep stages, where W is wakefulness, R is REM sleep, N1 is drowsiness (NREM1), N2 is light sleep (NREM2), and N3 is deep sleep (NREM3).

Here we can clearly see how a person starts to drift into sleep. At first, he enters the NREM1 stage, then “sinks” deeper (NREM2) till the deepest sleep stage (NREM3) at 11 pm (23:00). Then comes the fourth REM stage, shown in red on the graph.

Ok, if everything is clear with the sleep stages, then where are the sleep cycles? To answer this question, you need to know a simple rule:

Each sleep cycle always ends with a REM phase

There may be any order of sleep stages within a cycle (N1 -> N2 -> N1 -> W -> N2 -> N3), but the cycle always ends with a REM phase. Keeping this rule in mind, let’s take another look at the same hypnogram, but this time split into cycles (I’ll highlight each cycle with a distinct color for clarity):

sleep cycles on hypnogram

Here we see four sleep cycles with different phases within. During the first cycle, the stages were as follows: N1 -> N2 -> N3 -> REM; during the second cycle: N1 -> N2 -> N3 -> N2 -> W (awakening) -> N1 -> REM, and so on.

The brain takes a second round when one cycle ends with REM sleep. The REM stage always leads either to awakening (and then the first NREM1 sleep phase occurs) or to light sleep (NREM2):

sleep phases during one cycle

By the way, this is why the easiest and most comfortable awakening occurs only at the end of the cycle, when we either wake up on our own after the last REM stage or when the alarm clock awakens us at the beginning of the first phase (NREM1).

The worst awakening (with sleep inertia) occurs during the deep sleep stage when the brain is “off”. And the most exciting awakening is during REM sleep because this is when we see vivid dreams and can remember them if we interrupt this phase. What’s funny, even people who haven’t had dreams for years actually have at least 3 to 4 dreams every night.

Being aware of the 90-minute cycle rule, you can experiment with an alarm even without gadgets. For example, if you need to get up at 8 AM, try five cycles first (5 * 90 = 450 minutes or 7.5 hours, so 8 AM minus 7.5 hours = go to bed at 00:30 AM) and then six cycles (6 * 90 = 540 minutes or 9 hours, so 8 AM minus 9 hours = go to bed at 11:00 PM).

In other words, the point is to keep the alarm from ringing in the middle of the cycle, making you awake.

There is nothing complicated about sleep cycles, as you can see!

However, we haven’t yet figured out what these phases mean and why they come in a different order. But let’s pause at this point to answer the more fundamental question.

What’s going on with sleep anyway?

Indeed, why sleep has a cyclic structure, and why do the cycles consist of stages? Why are there four stages and not, say, twenty-four? What medical device can be used to obtain the same graphs and see if it’s true?

Actually, there is no such device. More precisely, a whole set of different sensors and devices are used. And all of them provide much harder to understand data:

polysomnography sample report

A specialist analyzes these waves and then draws the very same hypnograms that visually show how a person falls asleep.

But how do we step from these complicated charts to familiar sleep stages?

We have to answer this question to understand better the gap between the best fitness trackers or smartwatches and medical sleep analysis.

Each line on this graph represents data from a different device. For example, one sensor records body temperature (THERM line in the picture above), another one records blood oxygen levels (SpO2 line), the third one registers heart rate (HR line), two others track left and right eye movements (LEOG and REOG lines), and so on.

All of this must be taken into account to correctly and accurately detect sleep phases.

But the most important here are the first four lines:

eeg sample report

What do they mean? They are actually the key to understanding sleep cycles, as they tell us about brain rhythms.

Brain rhythms

A membrane isolates each cell, and there is always an electrical voltage across this membrane. It occurs due to ion channels and ion pumps which are special tiny mechanisms built into the cell membrane.

Some of these channels are constantly open so that charged ions (potassium, sodium, chlorine) can enter and leave the cell. Others open only after activation, which can be done either by voltage or chemically.

Charged ions can freely move inside and outside the cell, creating different charges on each side of the membrane. This charge difference is called membrane potential:

membrane potential

It’s an interesting topic, but we will not discuss it in detail here. The key thing is that there is a constant electric current flowing in the brain. And we can register it simply by placing wires (electrodes) on the head. Of course, it’s not as effective as putting the electrode directly into the brain, but it’s pretty handy.

The only problem is that this electrical signal is too weak. Moreover, if we put electrodes on the head of an actively awake person, the EEG shows us virtually nothing. This is because billions of neurons would randomly send signals to other cells, and we wouldn’t be able to pick out any particular signal from this noise..

But when a person calms down and closes their eyes, an amazing thing called neural synchronization happens. Tens and hundreds of thousands of neurons synchronize their activity and fire simultaneously. Such a powerful signal travels through the head and can be easily registered even outside.

These wave-like electricity fluctuations are observed between different groups of neurons with different frequencies and are called brain rhythms.

Thus, people have noticed that electrical waves of different frequencies constantly alternate during sleep. Rapid random oscillations with small amplitude during drowsiness turn to slow and powerful delta waves during deep sleep. Brain rhythms cyclically change while a person is asleep.

These cyclical changes in the brain’s electrical activity are called sleep phases or sleep stages. You can clearly see the difference in the electroencephalogram (EEG) depending on what phase a person is in:

brain waves description

Therefore, a specialist’s job is to correctly identify certain waves (bursts of the electrical activity of certain populations of neurons) and then draw a nice and clear sleep diagram. We can see certain features that make it easier to classify the waves: sleep spindles, K-complexes, theta waves, etc.

And at this point, you probably wondered, what do fitness trackers wondered, what do fitness trackers and smartwatches have to do with brain neurons and their electrical activity at all? Obviously, none. And that’s a huge problem!

So, how do fitness trackers and smartwatches track sleep phases?

As shown above, brain waves on EEG are only part of what has to be analyzed. It is impossible to get an accurate hypnogram without additional information (some features, like sleep spindles, may occur in different sleep stages).

This means that apart from waves, you also have to analyze body temperature, muscle movement, eye movement, breathing, heart rate, oxygen level, etc.

And here we already see familiar metrics from fitness trackers!

Initially, wearables only tracked muscle movements (using an accelerometer). As a result, they could, at best, more or less accurately detect bedtime and wake-up time.

Later on, a heart rate monitor (HRM) was added, and fitness trackers could track heart rate during sleep. HRM got more accurate, which made it possible to track heart rate variability.

Then fitness trackers got a pulse oximeter – a device that measures blood oxygen levels (SpO2). That allowed you to monitor your breathing (breath-holding caused by sleep apnea affects blood oxygen saturation). And heart rate variability was an additional source of data (when breathing in, the duration between successive heartbeats increases, and when breathing out, it decreases).

Today, manufacturers try to implement blood pressure and body temperature measurement features.

It turns out that fitness trackers and smartwatches track a lot of secondary indicators, except for the most important one – brain activity. Moreover, even if you could make an EEG and put the data into an application on your smartphone, the result would be noticeably different from the hypnogram made by a specialist. That’s because there is no fully automatic method of creating accurate hypnograms today.

Let’s see what happens during each phase of sleep to understand exactly how fitness trackers try to identify each sleep stage by indirect indicators.

What happens during the first phase of sleep (NREM1 or drowse)?

  • Heart rate decreases slightly.
  • Breathing slows down.
  • Eyes make slow movements.
  • Body temperature decreases.
  • Muscles relax with occasional twitches.

What happens during the second phase of sleep (NREM2 or light sleep)?

  • Heart rate and breathing become even slower.
  • Eye movements stop.
  • Muscles relax further.
  • Occasionally, heart rate and blood pressure may rise slightly for short periods.

What happens during the third phase of sleep (NREM3 or deep sleep)?

  • All physiological indicators (heart rate, respiration, body temperature) slow to their lowest levels.
  • Muscles are completely relaxed.
  • Brain activity is significantly reduced.

What happens during the fourth phase of sleep (REM)?

  • Your brain starts to work as actively as it does when you are awake. That’s when you have vivid dreams.
  • Your heart rate increases and becomes irregular. You experience intense emotions while sleeping.
  • Motor neurons are completely suppressed, causing your arms and legs to become paralyzed so that you don’t jump out of bed during sleep and get into trouble. If you wake up during REM sleep, you could experience sleep paralysis (you’re aware but unable to move or speak).
  • Your eyes move rapidly from side to side behind closed eyelids.

As you can see, a fitness tracker, in theory, is capable of detecting certain sleep stages with high accuracy. This is especially true for REM sleep when the accelerometer and gyroscope record the complete absence of movement while the heart rate monitor records a sudden HR increase.

However, in practice, things do not look so rosy. And now we will try to analyze several sleep graphs generated by fitness trackers and smartwatches.

How to analyze sleep phases with a fitness tracker

All fitness trackers not only track sleep differently but also draw different charts. Many trackers from Huawei, Samsung, Fitbit, Garmin, or Xiaomi draw classic hypnograms, just in a more colorful way. That is, we see a traditional sleep structure with “diving” into phases:

xiaomi watch sleep graph
sleep chart from samsung

But some manufacturers replace familiar hypnogram with much less visually clear graphs. In particular, a lot of trackers from Xiaomi and Huami draw illegible flat charts where sleep stages differ only in color:

zepp life sleep report
Zepp Life

As we already know, no smartwatch monitors the key measure – the brain’s electrical activity. Therefore, reports may be too far from reality, although sometimes some phases accurately match.

Due to this uncertainty/randomness, you should not take sleep analysis too seriously. Moreover, you can easily ignore reports like this since it clearly has a lot of mistakes:

faulty sleep report with a lot of mistakes
Huawei Watch 3

It’s impossible to identify separate cycles here, which, as we know, should end with the REM phase (shown in orange on the chart above). But here we see many REM stages, which is not true (unless you have narcolepsy), because you should typically have no more than 4-6 sleep cycles per night.

Besides, the sleep structure itself is unnatural here. We should see more deep sleep phases at the beginning and less at the end of the sleep. For REM sleep, the opposite is true.

There are other situations that should not be taken too seriously. For example, here we see a complete lack of REM sleep:

faulty sleep report without rem stage

Actually, it can happen, though there must be serious reasons for that, such as taking strong antidepressants.

Now let’s see a normal sleep chart, which you can analyze and trust:

quality sleep chart from samsung galaxy Watch 3
Galaxy Watch 3

Here we can easily identify separate sleep cycles, which are 4 in total:

sleep cycles on chart

We also see that most of the deep sleep was in the first half of the night, and the duration of the REM stage increased by morning. Overall, this is a pretty accurate hypnogram, except for minor errors. Most likely, one cycle is missing at the beginning since the first onset of the REM phase often occurs 70-110 minutes after falling asleep).

But how accurately this hypnogram reflects brain rhythms is another (and most important) question, which we will not answer unless we do a medical hypnogram at the same time.

There are more complicated cases, such as the one shown on this chart:

some sleep cycles can't be identified

First, we cannot precisely identify the length of the first sleep cycle since the REM phase began literally immediately and lasted almost the entire cycle, which is very unlikely. The first REM stage occurs 1-2 hours after falling asleep.

Also, the individual REM phases are split into smaller chunks:

poor quality sleep cycles analysis

Suppose you can identify separate cycles on your chart. In that case, you can pay attention to the total amount of sleep in each specific stage. However, if it’s hard to identify cycles or if there’s no phase on your chart, you should check if you’re wearing your fitness tracker correctly.

Okay, let’s assume you trust the fitness tracker’s readings and see credible reports. Then we have to deal with the last question.

Why do I need to monitor my stages of sleep?

It’s very important to go to bed at the same time to avoid circadian rhythm disorder (a frustrating phenomenon that can cause a lot of physical and psychological problems).

It’s also crucial to sleep enough time. But the sleep quality itself (the number of stages and their duration) plays an equally important role. You can sleep 7-8 hours and still have a lot of problems related to sleep if your sleep cycles are disrupted.

To be mentally and physically healthy, you should have all of the sleep stages in about this ratio (the table shows the percentage in relation to total sleep time):

Stage of sleepPercentage
Drowsiness (NREM1) and wakefulness2-5%
Light sleep (NREM2)45-60%
Deep sleep (NREM3)15-25%
REM sleep20-25%

It’s worth mentioning that certain changes occur in the sleep structure as we age. In particular, total sleep time as well as the percentage of REM and deep sleep decreases:

how sleep structure changes with age

If we take an average adult, the duration of deep sleep is directly correlated with the growth hormone secretion, which is vital:

growth hormone secretion

Also, there are many important physiological processes that take place during deep sleep, such as:

  • Consolidation of conscious memories and new information;
  • Clearing out waste products of neuronal metabolism accumulated in the brain during the day;
  • Removal of unnecessary connections between neurons;
  • Immune system recovery, etc.

REM phase is equally important, and its lack leads to various mental and emotional disorders (depressed mood, anger, apathy). In addition, many believe that it’s the REM phase that plays an important role in unconscious memory consolidation (new skills, etc.).

On the other hand, an excess of REM sleep during the night can have a negative effect. After all, the brain works during the REM stage as actively as it does during wakefulness; that is, it does not rest. And an excess of one phase is associated with a lack of another. In this case, you may feel broken, tired, and emotionally exhausted even after a long sleep.

Sleep stage analysis can also help identify various health issues. For example, if there are problems with breathing (sleep apnea), you’ll get a pretty pronounced hypnogram:

healthy vs sleep apnea hypnograms

The second hypnogram (sleep apnea) shows us a lack of deep and REM sleep caused by frequent awakenings. Once breathing stops, carbon dioxide accumulates in the blood, and the “dreaming” brain receives signals to turn on and fix the problem.

So, the deep sleep interrupts, the pharyngeal muscles tone increases, and the oxygen level is restored. After that, the brain goes back into “recovery mode” (deep sleep), but a few moments later, breathing problems arise again, and everything goes in a circle.

As a result, a person with such a hypnogram may sleep eight hours at a time, but their body will suffer from a lack of sleep anyway.

There are other problems that can only be noticed when analyzing sleep phases. For example, a disease of the nervous system called narcolepsy. In this case, the REM stage occurs literally immediately after falling asleep and is repeated very often (we saw something similar above on some charts).

Some medicines can inhibit certain sleep phases, and a person will not get the needed rest, even with a strict sleep schedule. The same goes for alcohol and nicotine. And all these problems can be noticed by analyzing sleep phases.

But let’s not forget!

I’ll repeat an important point once again. Fitness trackers and smartwatches can’t be reliable tools for sleep analysis. They can show total sleep time and its regularity with great accuracy, but when it comes to analyzing sleep quality, take it with a grain of salt!

Trackers do not measure the brain’s electrical activity in any way and try to identify sleep stages by secondary features.

If you want to improve your sleep quality with a fitness tracker, simply watch the tendencies when analyzing your sleep. Take the tracker readings as relative rather than absolute values, and watch them change over a period of time (whether the total amount of deep sleep or REM phase increases).

And most importantly, don’t worry if the tracker shows a lack or complete absence of some sleep phase. After all, it’s just the tracker’s “subjective opinion” and not a medical fact.

Alex Salo, Tech Longreads founder

If you enjoyed this post, please share it! Also, consider to support Tech Longreads on Patreon and get bonus materials!


References

  1. Physiology, Sleep Stages [Book]
  2. Hypnogram [source]
  3. Duration and Timing of Sleep are Associated with Repetitive Negative Thinking [DOI]
  4. Age-related changes in slow wave sleep and REM sleep and relationship with growth hormone and cortisol levels in healthy men [DOI]
  5. A role for REM sleep in recalibrating the sensitivity of the human brain to specific emotions [PMID: 20421251]
  6. Sleep [source]
4
4
Subscribe
Notify of

0 Comments
Inline Feedbacks
View all comments