Your smartwatch woke you up with a sleep score of 61 and a warning that your REM was low. Now you're anxious, caffeinated, and Googling sleep disorders before 8 a.m. Sound familiar? Before you spiral, it's worth understanding what that number actually means, how it was calculated, and what you should actually do with it.
Consumer wearables have gotten genuinely impressive. But there's a wide gap between "impressive" and "clinically accurate," and knowing where that gap sits will make you a smarter user of the data you're already collecting every night.
How Wearables Estimate Sleep (Without Reading Your Brain)
The gold standard for sleep analysis is polysomnography, a lab-based study that monitors brain electrical activity, eye movements, muscle tone, and breathing simultaneously. Your watch cannot do any of that. What it can do is measure a combination of signals that correlate with different sleep states.
Most modern smartwatches use three primary inputs:
- Heart rate variability (HRV): The fluctuation in time between heartbeats shifts meaningfully across sleep stages. Deep sleep and REM produce distinct HRV signatures that algorithms can partially distinguish.
- Accelerometry: Motion sensors detect when you're still, shifting, or restless. Near-total stillness combined with specific HRV patterns suggests deeper sleep stages.
- Blood oxygen saturation (SpO2): Dips in blood oxygen during sleep can indicate disrupted breathing, which is relevant to conditions like obstructive sleep apnea.
These signals, processed through proprietary machine learning algorithms, produce the sleep stage breakdowns and scores you see each morning. The critical word is estimate. Studies comparing consumer wearables to polysomnography consistently find that devices perform reasonably well at detecting total sleep time and distinguishing sleep from wakefulness, but are considerably less reliable when it comes to separating specific stages, particularly light sleep from REM.
Accuracy also varies significantly by device. A 2023 review comparing multiple consumer wearables found that agreement with polysomnography on REM staging ranged from poor to moderate across brands. The algorithm matters as much as the hardware, and both change with software updates, meaning your watch from two years ago and your watch today may classify the same night differently.
The Samsung Research That's Worth Paying Attention To
Not all wearable sleep research is equal, and a recent Samsung study deserves genuine attention rather than hype. Researchers used the Galaxy Watch's SpO2 and heart rate sensors to estimate the apnea-hypopnea index (AHI), the clinical metric used to grade obstructive sleep apnea (OSA) severity. The findings showed that the device's estimates correlated closely enough with lab-measured AHI to identify individuals at high risk of clinically significant apnea.
This matters because OSA is significantly underdiagnosed. Estimates suggest that roughly 80 to 90 percent of moderate-to-severe OSA cases in adults go undetected. The condition is linked to elevated cardiovascular risk, metabolic disruption, cognitive decline, and poor daytime function. A wearable that can flag high-risk individuals for clinical follow-up, without replacing the clinical test itself, is a meaningful public health tool.
What the research does not say is that your watch can diagnose sleep apnea. The study frames the technology as a screening signal, not a diagnostic instrument. If your device flags repeated oxygen dips or elevated respiratory disturbance estimates, the appropriate next step is talking to a physician, not self-diagnosing based on an app readout.
What Sleep Scores and REM Percentages Actually Mean
Every major platform, Apple, Garmin, Fitbit, Samsung, Oura, produces its own proprietary sleep score. These scores are not standardized across devices. A score of 75 on one platform does not mean the same thing as 75 on another, and neither is directly anchored to any clinical measure of sleep quality.
Sleep scores are composite metrics that typically weight factors like total sleep duration, estimated stage distribution, resting heart rate, and HRV. Each company decides the weighting. Each company updates its algorithm periodically. That means a score drop after a software update may reflect a change in math, not a change in your actual sleep.
REM percentage is similarly imprecise. Healthy adults typically spend 20 to 25 percent of sleep in REM, but that figure varies with age, alcohol intake, recent training load, medication use, and individual biology. A single night at 14 percent REM is not a crisis. It might reflect a hard training day, a late meal, or simply measurement error. Wearables tend to undercount REM in particular, so the number is almost certainly lower than your actual REM time.
HRV as a recovery proxy is arguably the most actionable of the common metrics, partly because it's measured with more precision than sleep staging, and partly because trends over time are more meaningful than any single reading. If your morning HRV is trending downward across two weeks of hard training, that's a signal worth responding to, ideally by prioritizing recovery and sleep consistency rather than optimizing your sleep score.
When to Pay Attention and When to Let It Go
The practical framework here is straightforward: use your wearable to identify patterns across weeks, not to evaluate any single night.
One poor sleep score tells you almost nothing diagnostically useful. It might mean you slept badly. It might mean you wore your watch too loosely, or moved around more than usual, or had a vivid dream that the accelerometer read as wakefulness. Single-night data is noisy.
What's worth taking seriously:
- Persistent low scores over two or more weeks that don't correlate with an obvious cause like travel, stress, or illness.
- Repeated SpO2 dips below 90 percent during sleep, which may indicate breathing disruption and warrant a conversation with your doctor.
- Consistently high resting heart rate during sleep relative to your personal baseline, which can signal inadequate recovery or early illness.
- A gradual downward trend in HRV across weeks, particularly if it coincides with fatigue, poor performance, or mood changes.
Lifestyle factors you can control have a larger documented effect on sleep quality than any optimization strategy your app will suggest. Consistent wake times, limiting alcohol, managing caffeine intake after midday, and keeping your sleep environment cool and dark all move the needle more reliably than chasing a number on a screen.
It's also worth noting that sleep doesn't exist in isolation from your broader daily habits. CBT-I Is the Insomnia Fix Most People Have Never Heard Of outlines the most evidence-supported behavioral approach to chronic sleep difficulty, and it works without a device in sight. If your scores are consistently poor and your daytime functioning is suffering, that framework is worth exploring long before you upgrade your watch.
The Orthosomnia Risk Is Real
Sleep researchers have identified a pattern they call orthosomnia: anxiety about achieving perfect sleep metrics that actually worsens sleep quality. If you're lying awake worrying about whether you'll get enough deep sleep to satisfy your app, the app is making things worse.
This is not a niche problem. As wearable adoption grows globally, more people are presenting to sleep clinics distressed about data from devices they don't fully understand. The data isn't the enemy, but treating a consumer algorithm as an authority on your health is a mistake. Your subjective experience of sleep, whether you feel rested, whether your energy is stable through the day, whether your mood and cognition are functioning well, remains more reliable than any single metric your watch produces.
Recovery doesn't happen in one dimension either. Everyday Movement Counts as Much as Your Workout is a useful reminder that how you spend your waking hours shapes your sleep just as much as what you do at bedtime. And if you're using supplements to support recovery, it's worth reading what the evidence actually says about energy and stamina claims before assuming a product is doing what the label promises.
The Bottom Line on Wearable Sleep Data
Your wearable is a useful tool for spotting broad trends, flagging potential issues that warrant medical attention, and building general awareness of how lifestyle choices affect your rest. It is not a sleep lab, it cannot diagnose conditions, and a bad score on a Tuesday morning is not a medical event.
Use the data as one input among many. If you're consistently waking up tired despite your watch reporting eight hours, that's a conversation for a doctor, not a cue to obsess over your sleep stage percentages. And if you find that checking your scores every morning makes you more anxious rather than more informed, it's completely reasonable to stop. The goal is better sleep, not a better number.
As screening tools improve, as the Samsung research suggests they will, wearables may play a meaningful clinical role in identifying conditions like OSA earlier than current pathways allow. That's genuinely promising. But right now, the most valuable thing your watch can do is point you in a direction, not tell you what's wrong.